CN119722293A - Information processing system, method, device and medium based on bank loan platform - Google Patents

Abstract

The present invention relates to the field of information processing technology, and discloses an information processing system, method, device and medium based on a bank loan platform. The system includes: a data embedding design module, which is used to determine the data embedding plan according to the loan application process and user behavior characteristics; a data acquisition module, which is used to collect user data in real time according to the data embedding plan, and send the user data to the back-end server; a data processing and analysis module, which is used to receive and process and analyze user data to determine the data analysis results; the result application module includes a loan approval optimization unit, which is used to optimize the loan approval process according to the data analysis results, including: adjusting the loan amount and interest rate, making loan approval decisions, so as to speed up the loan approval process. The present invention can collect user data of users in the loan application process in real time, analyze the user data to obtain analysis results, thereby adjusting the loan amount and interest rate, making loan approval decisions, and speeding up the loan approval process.

Description

Information processing system, method, equipment and medium based on bank loan platform

Technical Field

The invention relates to the technical field of information processing, in particular to an information processing system, method, equipment and medium based on a bank loan platform.

Background

With the rapid development of digital finance technology, online loans become an important trend, and all steps of loan application can be completed without going out, including knowing the application conditions of various loans, preparing application data, and can be efficiently completed on the Internet until the loan application is submitted. The user submits the application on line, and the consumer finance company performs preliminary screening and auditing, and for the application with higher part of risks or needing further verification, links such as face notes, home visits and the like under the line can be arranged so as to evaluate the risks more accurately. In the above-mentioned on-line application off-line auditing method, in some auditing processes, the manual intervention links are more, and the method is easily influenced by subjective factors, such as experience, emotion, prejudice, etc. of the approver, thereby influencing objectivity and fairness of the auditing result, and the auditing efficiency of the manual auditing method is lower. Therefore, there is a need for an information processing scheme that can improve the accuracy and efficiency of loan approval.

Disclosure of Invention

Therefore, the invention aims to solve the technical problems of lower manual auditing accuracy and lower efficiency in the related technology.

In order to solve the above technical problems, the present invention provides an information processing system based on a bank loan platform, comprising:

The data embedded point design module is used for determining a data embedded point scheme according to the business process and the user behavior characteristics of the front end H5 page of the bank loan platform, wherein the business process comprises a loan application process;

the data acquisition module is configured in the front-end H5 page and is used for acquiring user data in real time according to the data embedding scheme determined by the data embedding design module and transmitting the user data to the rear-end server;

The user data comprises user behavior data and application information, wherein the behavior data comprises a browsing path, a stay time and a clicking event of the user, the application information comprises application time, loan amount, loan period, repayment mode, personal information, enterprise operation information, enterprise scale and operation information, credit history, historical loan records and historical repayment records;

the data processing and analyzing module is configured at the back-end server and is used for receiving and processing and analyzing the user data to determine a data analysis result;

The result application module comprises a loan approval optimizing unit, and is used for optimizing a loan approval process according to the data analysis result, wherein the loan approval optimizing unit is used for adjusting the loan amount and the interest rate and making a loan approval decision so as to accelerate the loan approval process.

In an alternative embodiment, the data embedding point design module includes:

The embedded point demand determining unit is used for determining the embedded point demand according to the business flow and the user behavior characteristics of the front end H5 page of the bank loan platform, wherein the embedded point demand comprises key nodes and data fields corresponding to the key nodes;

the embedded point scheme design unit is used for determining embedded point positions according to the key nodes in the embedded point requirements, determining data types according to the data fields corresponding to the key nodes, and designing corresponding acquisition frequencies.

In an alternative embodiment, the data acquisition module includes:

The system comprises a data embedding point design module, a data acquisition unit, a data encryption unit, a data compression unit and a data transmission unit, wherein the data embedding point design module is used for determining a data embedding point scheme of the data embedding point design module, the data acquisition unit is used for acquiring user data in real time according to the data embedding point scheme determined by the data embedding point design module, the data encryption unit is used for conducting encryption processing on the user data to obtain encrypted user data, the data compression unit is used for compressing the encrypted user data to obtain compressed user data, and the data transmission unit is used for sending the compressed user data to the back-end server.

In an alternative embodiment, the data processing and analysis module includes:

the data processing unit is used for receiving the user data and carrying out data processing on the user data to obtain user data after data processing, wherein the data processing comprises preprocessing and feature extraction, and the data analysis unit is used for analyzing the user data after data processing to obtain the data analysis result.

In an alternative embodiment, the data analysis unit comprises:

the model building unit is used for building a risk assessment model;

the model application unit is used for inputting the user data after data processing into the risk assessment model as input data to obtain risk scores corresponding to the users, and the risk scores are used for representing credit risks and loan risks of the users.

In an optional implementation manner, the loan approval optimizing unit is specifically configured to adjust the loan amount and the interest rate according to the risk score corresponding to the user, and automatically or manually make a loan approval decision according to the risk score corresponding to the user and a preset approval standard.

In an alternative embodiment, the data analysis unit further comprises:

And the user portrait construction unit is used for analyzing the user data after the data processing to construct a user portrait.

In a second aspect, the present invention provides an information processing method based on a bank loan platform, including:

Determining a data embedded point scheme according to the business process of the front end H5 page of the bank loan platform and the behavior characteristics of a user, wherein the business process comprises a loan application process;

the method comprises the steps of collecting user data in real time according to a data burial point scheme and sending the user data to a back-end server, wherein the user data comprises behavior data of a user and application information, the behavior data comprises a browsing path, stay time and clicking events of the user, the application information comprises application time, loan amount, loan deadline, repayment mode and personal information, name, age and sex of the user, enterprise operation information comprises enterprise scale and operation information, and credit history comprises a history loan record and a history repayment record;

receiving and processing and analyzing the user data to determine a data analysis result;

optimizing the loan approval process according to the data analysis result comprises adjusting the loan amount and the interest rate, and making a loan approval decision so as to accelerate the loan approval process.

In a third aspect, the present invention provides a computer device comprising a memory and a processor, the memory and the processor being communicatively coupled to each other, the memory having stored therein computer instructions, the processor executing the computer instructions to thereby perform the banking platform based information processing method of the second aspect.

In a fourth aspect, the present invention provides a computer-readable storage medium having stored thereon computer instructions for causing a computer to execute the information processing method based on the bank loan platform of any of the embodiments of the second aspect.

In a fifth aspect, the present invention provides a computer program product comprising computer instructions for causing a computer to perform the information processing method of the second aspect described above.

The technical scheme provided by the invention has the following technical effects:

According to the technical scheme provided by the embodiment of the invention, the data embedded point scheme can be determined according to the business flow and the user behavior characteristics of the H5 page at the front end of the bank loan platform through the data embedded point design module. This enables the accurate collection of various data of the user in the loan application process, including personal information (personal information, business administration information, credit history), behavioral data (browsing path, stay time, click event), and device information (device model, operating system, network environment). Compared with manual auditing which may depend on limited data and subjective judgment, the data acquisition mode is more comprehensive and objective, and provides a rich data basis for accurate evaluation.

The data acquisition module can acquire user data in real time and send the user data to the back-end server. The method means that the timeliness of the data is high, the problems that delay and information update are not timely in the process of manually collecting the data are avoided, the data processing and analyzing module at the rear end can analyze based on the latest data, and the accuracy of analysis results is improved.

The data processing and analyzing module receives and processes and analyzes the user data at the back-end server to quickly obtain a data analysis result. The process avoids the tedious process of manually checking and analyzing the user data one by one, and greatly improves the efficiency.

The loan approval optimizing unit in the result application module optimizes the loan approval process according to the data analysis result, such as adjusting the loan amount and the interest rate, and making a loan approval decision. The decision-making mode based on data driving can accelerate the loan approval process, reduce links such as repeated communication, data supplementation and the like possibly occurring in the manual approval process, and improve the overall approval efficiency.

Through comprehensive and real-time data acquisition and automatic data analysis, more accurate information can be provided for loan approval, inaccuracy caused by factors such as limited information, subjective judgment and the like during manual auditing is reduced, and therefore the problem of lower accuracy of the manual auditing loan is effectively solved. The automatic data processing and the data-based approval process optimization reduce manual intervention links, quicken the approval process and solve the problem of lower manual auditing efficiency. Therefore, the technical scheme of the invention can solve the technical problems of lower accuracy and lower efficiency of manual auditing loan to a certain extent.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the related art, the drawings that are required to be used in the description of the embodiments or the related art will be briefly described, and it is apparent that the drawings in the description below are some embodiments of the present invention, and other drawings may be obtained according to the drawings without inventive effort for those skilled in the art.

FIG. 1 is a schematic diagram of a bank loan platform-based information processing system, according to an embodiment of the invention;

FIG. 2 is a schematic diagram of a data embedding scheme according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a visual analysis result according to an embodiment of the present invention;

FIG. 4 is a flowchart of an information processing method based on a bank loan platform, according to an embodiment of the invention;

fig. 5 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

Therefore, the embodiment of the invention provides an information processing system, method, equipment and medium based on a bank loan platform, which are used for solving the problems of lower accuracy and lower efficiency of manual auditing.

According to an embodiment of the present invention, an embodiment of an information processing system based on a bank loan platform is provided, and it should be noted that a single system is used to implement the following embodiments and alternative implementations. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. While the system described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.

FIG. 1 is a schematic diagram of another information processing system based on a loan platform of the invention. As shown in fig. 1, in an embodiment of the present invention, there is provided an information processing system based on a loan platform, including:

The system comprises a data embedded point design module 11, a data acquisition module 12, a data processing and analyzing module 13 and a result application module 14.

The data buried point design module 11 includes a buried point demand determination unit 111 and a buried point scheme design unit 112.

The data acquisition module 12 includes a data acquisition unit 121, a data encryption unit 122, a data compression unit 123, and a data transmission unit 124.

The data processing and analyzing module 13 includes a data processing unit 131, a data analyzing unit 132. The data analysis unit 132 includes a model construction unit 1321 and a model application unit 1322. The data analysis unit 132 further includes a user portrayal construction unit 1323.

The results application module 14 includes a loan approval optimization unit 141.

The data embedded point design module 11 is used for determining a data embedded point scheme according to the business process and the user behavior characteristics of the front end H5 page of the bank loan platform. The business process includes a loan application process. The business processes can also include a product browsing and comparing process, a user authentication and authorization process, a consultation and customer service process, a repayment management process, and the like. As an example, a data embedding scheme is shown in fig. 2.

Product browsing and comparing processes:

Product display page browsing, namely when a user enters a loan product display page, recording the type of loan products (such as personal consumption loans, housing loans, enterprise business loans and the like) browsed by the user and the browsing duration (residence time) of each product. This may help to learn the initial interests of the user for different loan products. For example, a user staying longer on a home loan product page may mean that he has a potential need to purchase a home loan.

And checking the product details, namely burying points for the action of clicking and checking the loan product details by a user, and recording the checked product detail contents including information such as interest rate, loan deadline, repayment mode, limit range and the like. The data can analyze the loan product characteristics focused by the user, and provide basis for accurate marketing and product optimization. For example, if many users frequently see if advance payouts of a product have terms for a default, a clearer explanation of this in the product introduction may be required.

Product comparison behavior when the user uses the product comparison function (if any) provided by the platform, record the product combinations being compared and the dimensions of the comparison (e.g., interest rate, line flexibility, etc.). This helps to understand the trade-off factors of the user in selecting loan products in order to adjust the product policy.

2. User authentication and authorization flow:

And the identity authentication step is to record the completion condition of each step, whether errors occur or not or repeat operation in the steps of uploading an identity card photo, inputting a mobile phone number to obtain a verification code, identifying face (if any) and the like in the identity authentication link of the user. These data are important for optimizing authentication procedures, improving user experience, and protecting against authentication risks. For example, if many users fail multiple times in the face recognition step, it may be necessary to check whether the function has technical problems or to provide more explicit operation guidance.

And the authorization operation, namely inquiring credit information of the user authorization platform, and burying points for bank running water and other actions, including whether the authorization is successful or not, the authorization time point and the like. Knowing the user's acceptance of authorization and possible problems helps to improve authorization alerts and descriptions, while also better tracking the user's process of credit data acquisition.

3. Consultation and customer service flow:

The online consultation is initiated by recording the entrance position (such as a product detail page, an application flow or a special customer service page) of the online consultation initiated by the user, and the problem type (such as product detail, application condition, interest rate calculation and the like) of the consultation. This may help to learn the user's confusion points during the loan process in order to optimize the product introduction and customer service knowledge base. Customer service communication details, namely if the platform has a chat type customer service function, recording chat contents, chat duration, problem solving conditions and the like of customer service and users. These data help to evaluate customer service quality while also mining user common problems for optimizing platform common problem solving (FAQ) boards.

Customer service evaluation feedback, namely burying points for the evaluation of the customer service by the user, wherein the points comprise evaluation grades (such as satisfaction, general dissatisfaction) and specific evaluation contents. This is important for improving customer service quality and user satisfaction.

4. Repayment management flow (for loaned user):

Repayment plan viewing, in which when a loaned user logs in to the platform to view the repayment plan, the frequency of viewing, the specific time point of viewing (such as the time near the repayment date, a long time in advance, etc.), and whether there is an operation to print or download the repayment plan are recorded. This may help to learn the user's attention to the payment plan and management habits.

And the repayment reminding interaction is to record whether a user views the reminding and the operation after viewing (such as neglecting, clicking to enter a repayment page and the like) for repayment reminding information (short messages, push notifications and the like) sent by the platform. These data help optimize the payoff alert strategy and increase payoff rate.

And the repayment operation records the details of the repayment mode, whether the repayment is successful, whether the repayment is advanced and the like when the user performs repayment operations (such as online payment repayment, automatic repayment setting and the like). This is important for assessing the repayment capabilities and credit status of the user.

In this embodiment, the data embedding point design module 11 may include:

The embedded point demand determining unit 111 is configured to determine an embedded point demand according to a business process and a user behavior characteristic of a front end H5 page of a bank loan platform. The embedded point requirement comprises a key node and a data field corresponding to the key node.

The buried point scheme design unit 112 is configured to determine a buried point location according to a key node in the buried point requirement, determine a data type according to a data field corresponding to the key node, and design a corresponding acquisition frequency.

The construction of flow data in the traditional technology is dilemma:

poor quality of buried points, misburied and missed buried, uneven quality and risk of using data.

Buried point design is irregular, so that buried points are scattered and disordered, unified use is not facilitated, various data lack of consent management, and data value is difficult to develop.

The service utilization rate is low, and the cost is consumed when the service is not used.

Previous pain points of various roles of banking platform in traffic data practice:

The buried points are not uniformly managed by the data users (business departments). Logic has no meaning, attribute content and responsible person are absent.

Buried point developers have no record of new, changed, abandoned, offline and the like buried points, and past buried point codes are not dared to move, are more and more chaotic and the like.

The embedded point demand party (the operator of the bank loan platform, the front, middle and back team) has no maintenance, and flexible and controlled version management can not be performed.

The data manager (the operator of the bank loan platform) lacks management tools for the current status of flow data assets, the number of buried points, the data volume, the cost consumption, the number of effective buried points and the like.

The embedded point demand determining unit is specifically used for disassembling the loan application process, wherein the loan application process is subdivided into a plurality of sub-processes, such as the stages of registering and logging in by a user, filling in personal information (personal information, enterprise information and the like), uploading data (identity card photos, business licenses and the like), selecting loan product types and amounts, submitting application and auditing and the like.

The interactive behavior of the user in each sub-flow is analyzed, such as the input behavior of the user, the operation of the necessary and optional filling items in the stage of filling the personal information. And when uploading data, file selection, feedback processing of success or failure of uploading and the like.

The key behavior determination is to find out user behaviors which have important influence on loan approval, such as browsing duration of detail pages of loan products (reflecting the attention and understanding degree of the users to the products), the number of modification times (possibly implying the authenticity and stability of information) when credit history information is filled in, the number of repeated views of the whole application page (embodying the cautiousness degree of the users) before submitting the application, and the like.

And determining the buried point requirement, namely determining the buried point key node based on the business process disassembly and key behavior analysis. Such as a login page, a login button click, a login success event, etc. The personal information fills in the page, namely, the fields of name, age, gender and the like begin to input the event, complete the input event and save the information event. Enterprise business information page, enterprise scale field input, business information text box content change event, etc. Credit history page, history loan record inquiry button click, history repayment record display event, etc. Loan product selection page, product click event, limit selection event, etc. Submitting an application page, namely submitting a button click event, applying for a successful submitting event and the like.

Corresponding data field determination, namely determining the data field to be acquired for each key node. For example, a login page, a user name, a password, a login time, a login IP address, etc. The personal information fills in the page including name, age, sex, ID card number, contact information, etc. Enterprise business information page, enterprise scale (concrete data such as employee number, registered capital, etc.), business scope (text description), enterprise establishment time, etc. Credit history page, history loan amount, loan period, repayment mode, repayment time, whether overdue record exists, etc. Loan product selection page, product name, product number, selected credit, interest rate type, etc. The application page is submitted, and the application time, the application form number, the application state and the like are applied.

The embedded point scheme design unit is specifically used for determining embedded point positions, data types and acquisition frequencies.

Burial point location determination for front end H5 pages, burial points are typically found in a JavaScript function triggered by a related event. For example, a click event of the login button buries a point, and a data acquisition code is inserted into a click response function of the login button. And for data acquisition when the page loading is completed (such as equipment information is acquired by first loading of the page), embedding points in the 'onload' event function of the page.

Data type determination, namely determining the data type according to the property of the data field. For example, the text information such as name, business name, etc. is a character string type. Age, number of employees, loan amount, etc. are numerical types. Registration time, login time, repayment time, etc. are date-time types. The device model, the operating system and the like are character string types (the enumeration values of the device type and the operating system can be predefined, and subsequent analysis is convenient). Whether there is an overdue record is of the boolean type (there is an overdue true, and there is no overdue false).

The acquisition frequency is determined, namely, for some disposable events, such as successful registration, successful application submission and the like, the acquisition frequency is 1 time, namely, the events are acquired when the events occur. For continuous behavior data of the user on the page, such as a browsing path, a higher acquisition frequency can be set, for example, URL information of the current page of the user is acquired every 1 second, so that the browsing track of the user is completely recorded. For the content change of the user information input field, the content change can be collected when the content is changed each time, and the real-time performance and the integrity of the data are ensured. For the equipment information, the user can acquire the equipment information 1 time when the page is loaded because the equipment information is basically determined and does not change frequently when entering the page.

The embedded point is the home page, and the button is clicked, such as an 'immediate application' button, an 'know more' button and the like, and the entering intention of the user is recorded. Page dwell time-the dwell time of the user on the home page is recorded to evaluate the home page appeal. And applying for a flow page, namely filling a form, wherein the input event of each input box comprises input content, input time and the like, and the smoothness of filling and the possible problems of a user are known. And the drop-down menu is selected, and the selection condition of the user when the drop-down menu such as loan deadline, repayment mode and the like is selected is recorded. And the submitting button records the time and the times of the user submitting the application. And (5) a result page, namely, approval result display, namely, recording the approval result (passing, refusal and the like) seen by the user and the time for viewing the result. The button for guiding the next operation, such as "reapply", "view details", etc.

Data type 1) behavior data A, clicking event A, recording elements such as buttons, links and the like clicked by a user. B. Page view event-recording the time when the user entered and exited each page. C. Form fill-in data including entered content, selected options, etc. 2) And the state data is A, the network state is recorded, and the network conditions of the user in the operation process are good, poor and the like. 3) Device information including device type, operating system version, etc. 4) Business data, information such as loan application amount, term, purpose and the like, and 5) approval result status (approval passing and non-passing).

The acquisition frequency is 1) the important behaviors such as button clicking events, page switching events and the like are acquired in real time, so that the user operation can be known in time. 2) The time acquisition is that the page stay time can be acquired once at intervals (such as 10 seconds), and finally the total stay time is calculated. During the form filling process, the input state can be collected at intervals (such as 30 seconds) to prevent data loss.

The data acquisition module 12 is configured in the front end H5 page, and is configured to acquire user data in real time according to the data embedding scheme determined by the data embedding design module, and send the user data to the back end server.

In this embodiment, after the user data is sent to the back-end server, data filtering may be performed on the user data collected in real time, and the user data satisfying the filtering condition may be filtered by inputting the filtering condition.

In the present embodiment, the user data includes, but is not limited to, behavior data and application information of the user. The behavior data includes a browsing path, a stay time, and a click event of the user. The application information comprises application time, loan amount, loan period, repayment mode, personal information, such as the name, age and sex of the user, enterprise operation information, financial conditions such as enterprise scale and operation information, credit history, historical loan record and historical repayment record.

And designing an index system for data embedded point acquisition, wherein the index system comprises personal information, behavior data, equipment information and the like of a user. The personal information may include, among other things, the name, age, gender, etc. of the user. The behavior data may include a user's browsing path, dwell time, click event, etc. The device information may include a user's device model, operating system, network environment, etc. By comprehensively collecting the indexes, the comprehensiveness and accuracy of the data are ensured.

In this embodiment, the user data also includes device information including the user's device model, operating system, and network environment.

In this embodiment, the data acquisition module 12 may include:

The data acquisition unit 121 is configured to acquire user data in real time according to the data embedding scheme determined by the data embedding design module.

In this embodiment, the data acquisition unit 121 is specifically configured to acquire data at the buried point according to the corresponding acquisition frequency and data type.

The data encryption unit 122 is configured to encrypt the user data to obtain encrypted user data.

In this embodiment, the encrypted user data may be obtained by encrypting the user data by a conventional technical means in the art. Such as symmetric encryption algorithms, asymmetric encryption algorithms, hash encryption, etc. As one example, the encryption process is performed using the AES algorithm, which determines the key length, the AES supports 128-bit, 192-bit and 256-bit keys, the longer the key length, the higher the security, but the encryption and decryption may be slower. For example, selecting a 128 bit key may generate a key of 16 bytes in length (128 bits converted to bytes) by a secure random number generator. Encryption process after the user data is obtained at the front-end H5 page or at the back-end server (depending on the specific encryption implementation location), the data is encrypted using the selected AES algorithm and key. For example, in a JavaScript environment (front-end), it may be implemented with some sophisticated encryption libraries (e.g., cryptoJS). After the backend server receives the encrypted user data (assuming it is transmitted in Base64 encoded form), it needs to perform Base64 decoding first and then decrypt it using the same key.

The data compression unit 123 is configured to compress the encrypted user data to obtain compressed user data.

In this embodiment, the encrypted user data may be compressed by a technical means conventional in the art, to obtain compressed user data. Such as lossless compression algorithm, GZIP algorithm.

And a data transmission unit 124, configured to send the compressed user data to the backend server.

In this embodiment, the compressed user data may be sent to the backend server by using a technical means that is conventional in the art. For example, HTTP protocol, message queue transmission, etc. are used.

In the embodiment, in the front end H5 page of the bank loan platform, the data embedded point SDK is introduced to realize real-time acquisition of user behaviors. The data embedded point SDK can collect behavior data such as clicking, inputting, sliding and the like of a user on a page and send the behavior data to a back-end server for processing in real time.

And the data processing and analyzing module 13 is configured at the back-end server and is used for receiving and processing and analyzing the user data to determine a data analysis result.

In this embodiment, the data processing and analyzing module 13 may include:

The data processing unit 131 is configured to receive user data, and perform data processing on the user data to obtain user data after data processing. The data processing comprises preprocessing and feature extraction. Preprocessing includes data cleansing, removal of invalid, erroneous or duplicate data, formatting and normalization. And cleaning the collected user data to remove invalid, wrong or repeated data. For example, if the user age field presents a significantly unreasonable value (e.g., over 150 years), then the record is revised or deleted. The user data is formatted and normalized to make it suitable for subsequent analysis. Feature extraction, namely extracting features from user data, such as browsing paths, operating frequency, normalization of input content and the like of the user, and constructing a more comprehensive feature set by combining application information of the user, such as basic information, financial data and the like. Features that have a significant impact on risk assessment, such as age, income level in personal information, overdue times in credit history, historical loan times, recent loan product browsing frequency in behavioral data, etc., are selected from the many-user data. At the same time, some features may be combined or new features derived, such as combining age and income level to generate a comprehensive index reflecting compensation. For classification features (such as gender, occupation, etc.), coding and conversion into numerical form are needed for model processing, and common coding methods include single-hot coding. For numerical features, in order to make different features in the same dimension and facilitate model learning, normalization processing may be performed, such as mapping the numerical value to a 0-1 interval or performing normalization processing (making the mean value of the numerical value be 0 and the variance be 1).

The data analysis unit 132 is configured to analyze the user data after the data processing, and obtain a data analysis result.

In the present embodiment, the data analysis unit 132 may include a model construction unit 1321 and a model application unit 1322.

The model building unit 1321 is configured to build a data analysis model, where the data analysis model includes a risk assessment model, and further includes a fraud detection model, an approval efficiency prediction model, and the like.

In this embodiment, the data analysis model may be constructed using techniques such as machine learning, data mining, and the like. Data analysis models include, but are not limited to, risk assessment models, fraud detection models, approval efficiency prediction models, and the like.

And (3) carrying out real-time data analysis through a data analysis model, namely carrying out rapid analysis on the data transmitted in real time, and calculating key indexes such as user risk scores, fraud risks and the like. Bottlenecks and potential problems in the approval process are analyzed, such as overlong time consumption of certain links, high information filling error rate of users and the like.

And the strategy can be dynamically adjusted through the result application module, namely the approval process and strategy are dynamically adjusted according to the data analysis result. For example, for users with high risk scores, the approval process can be simplified, and the approval time can be shortened. For applications with high fraud risk, the auditing strength can be enhanced or the application can be denied.

For the risk assessment model, in this embodiment, the user data after the data processing is divided into a training set, a verification set and a test set according to a certain proportion. Typically, a training set is used to train the model so that the model learns the relationship between the features and the risk. The verification set is used for adjusting the super parameters of the model in the model training process and selecting the optimal model configuration. The test set is used to ultimately evaluate the performance of the model on unseen data, and a common division ratio may be 7:2:1 (training set: validation set: test set). Training with training set taking logistic regression model as an example, training with partitioned training set data (assuming feature data is stored in x_train matrix and corresponding risk tag data is stored in y_train vector, risk tag may be 0 to represent low risk and 1 to represent high risk). Hyper-parametric tuning (optional) for some models (such as parameters of maximum depth of decision tree, number of trees of random forest, etc.), hyper-parametric tuning is required by a validation set to find the optimal combination of parameters to optimize model performance. Common hyper-parameter adjustment methods are grid search (GRIDSEARCH), random search (RandomSearch), and the like. Model evaluation, namely evaluating the trained model by using test set data (characteristic data is X_test, corresponding risk label is y_test), wherein common evaluation indexes are Accuracy (Accumey), precision (Precision), recall (Recall), F1 value, area under ROC curve (AUC) and the like. The risk score acquisition and application are that user data (after the same characteristic engineering treatment is carried out) which is needed to evaluate risks in the practical application is taken as input into a trained risk evaluation model, the model outputs corresponding prediction results (for a classification model, such as 0 or 1, a probability value between 0 and 1, which can be output as a probability, can be taken as a risk score, and the probability is higher, the risk of credit and loan risk of a user are characterized and decided according to the risk score, for example, a threshold (such as 0.5) is set, the user with the risk score being larger than the threshold is regarded as a high-risk user, and more cautious measures (such as reducing the amount of money, improving the interest rate or directly rejecting loan application and the like) can be taken in loan approval.

The model application unit 1322 is configured to input the user data after the data processing as input data to the risk assessment model to obtain a risk score corresponding to the user. The risk score is used to characterize the credit risk and loan risk of the user.

In this embodiment, the risk score calculation includes:

user data collection, which collects multidimensional data of users, including but not limited to personal information (e.g., age, occupation, income, etc.), business management information (e.g., business scale, revenue, profitability, etc.), credit history (e.g., past loan records, repayment, etc.), behavioral data (e.g., operational behavior on H5 pages, residence time, etc.).

And (4) setting indexes, namely determining index weights corresponding to various data. For example, credit history may be a relatively high weight because it directly reflects the user's repayment capabilities and credit status. The enterprise management information also has a certain weight, so that the economical strength and stability of the user are reflected.

And (3) scoring the model, namely comprehensively calculating each index by using a mathematical model (risk assessment model). Common models are linear regression, logistic regression, decision trees, etc. that can be used as risk assessment models. Values of different indicators are translated into a specific risk score by these models.

In this embodiment, the risk assessment model is built by training the risk assessment model based on user data using a machine learning algorithm. The risk assessment model learns how to predict a user's loan risk based on the features.

And (3) analyzing in real time, namely when the user operates on the H5 page, processing the user data and the behavior data in real time by the server, performing risk assessment by using a trained model, and outputting a risk score by the model to reflect the credit risk and loan risk of the user.

In this embodiment, the data analysis unit 132 may further include a user portrait construction unit 1323 for analyzing the user data after the data processing to construct a user portrait.

The following are the detailed steps of constructing a user representation:

1. Data collection and integration:

1. Personal information collection, namely, age, obtained from user registration information or real-name authentication channels. Different age phases often have different consumption and loan requirements, for example, a young person may prefer to consume a loan for educational promotion or purchasing of electronic products, while a middle-aged person may consider a housing loan or a commercial investment loan more.

And the occupation is completed by filling in work units and position information by the user or supplementing and perfecting the occupation information by a third-party data platform. Occupation is closely related to income stability, loan use and repayment capability, such as public officers and large enterprise staff, usually have a relatively stable income source, and can have certain advantages in loan approval. And the income of free professionals or entrepreneurs is greatly fluctuated, and the risk assessment needs to integrate multiple factors.

Revenue can be actively declared by the user but needs to be verified in combination with other data. Some banking platforms may require users to provide payroll, tax proof, etc. to more accurately determine revenue levels to assess their loan compensation capabilities and credit risk. For example, a high-revenue user may have a higher credit demand and a greater repayment capacity, but may also face more investment temptation and potential risks.

2. Enterprise business information collection (for the enterprise owner or related user) enterprise size is determined based on the information of the number of employees, the asset size, the office area, etc. of the enterprise. Large enterprises may have advantages in financing channels and costs, and their loan requirements often relate to strategic decisions such as expansion, and purchase. While small micro-enterprises may pay more attention to mobile fund loans to maintain daily operation, and because of smaller scale, the risk resistance is relatively weak, and in risk assessment, the operational stability and cash flow conditions need to be focused. And acquiring data through enterprise financial report forms, tax declaration records or docking with enterprise financial software. Enterprises with stable revenue growth and good profitability are generally regarded as low-risk clients when applying loans, and more favorable loan conditions can be obtained. Conversely, an enterprise with a smooth or long-term loss of revenue may face a more rigorous review of its loan application, and the bank may require additional guaranties or mortgages to be provided.

3. The credit history collection past loan records go on an expedition the credit system inquires all the historical loan information of the user, including loan organization, loan amount, loan period, repayment state and the like. The risk score of the user who pays for many times on time and full amount is higher, and the credit risk is lower. However, the user with overdue records, especially the user with more overdue times or longer overdue time, has a significantly increased credit risk, and the bank may increase the interest rate, decrease the credit limit or directly reject the application when approving the loan. And (3) analyzing repayment details of each loan in detail, and if the repayment details are early repayment, partial repayment, overdue repayment frequency and time length and the like. Advanced payouts may reflect that the user is more funded or sensitive to interest rate. And frequent overdue repayment indicates that the repayment capability or repayment willingness of the user is problematic, and is an important negative index for credit risk assessment.

4. Behavior data collection (based on H5 page) operation behavior, recording operations such as clicking, sliding, submitting and the like of a user on the H5 page. For example, a user frequently clicking on a high-value loan product page may suggest that it has a greater funding demand. And the detailed pages of the loan products are deeply browsed, including information such as interest rate, repayment mode, deadline and the like, so that the user has higher attention and understanding intention on the loan products, and related information can be provided pertinently or audit can be enhanced in subsequent marketing or risk assessment. Dwell time, which is to count the dwell time of the user on different pages or specific functional modules. Stay on the credit assessment prompt page for a long time may indicate that the user is more concerned about his credit status or is carefully checking the information. The user may not submit the loan application, but may be hesitant to the loan decision, and need to further analyze the reasons, such as whether there is difficulty in filling information, doubt about the terms of the loan, etc., so as to provide corresponding help or optimize the page design.

2. Data cleaning and pretreatment:

1. the missing value processing is to judge the cause of the missing and the importance of the data for the missing data in the personal information, the enterprise management information or the credit history. If the key data is missing (such as missing revenue information and cannot be verified by other ways), further communication with the user may be required to obtain the supplemental information, or reasonable speculation or population may be performed based on other relevant information of the user. For example, for users with a long working life who are professional teachers, the average income level of the same-area co-office teachers can be referenced for filling. For non-critical data misses (e.g., a detailed house number miss in an enterprise registry), mode filling (if there are more identical values in this field) or direct ignoring of the missing value may be used, depending on the context of the data usage and the impact on model accuracy.

2. Outlier handling examines outliers in personal information such as age out of reasonable range (over 120 years or less than 0 years), income too high or too low (severely inconsistent with average levels in profession or territory), etc. For outliers, further verification of their authenticity is required. If the user is misfilled or data is input wrong, the user is corrected in time. If special cases (such as high-income stars or enterprise high-rise), relevant proving materials need to be collected for confirmation. In enterprise management information, such as abnormal large fluctuation of revenue or profit, the reasons behind the enterprise need to be deeply analyzed, which may be that the enterprise carries out great strategic adjustment, market sudden change or financial data fake making and the like, and the reliability of the data is ensured by comparing and verifying the enterprise financial statement, industry report and other multi-channel data. For the abnormal overdue records in the credit history (for example, overdue time of a loan is up to several years and is not processed, but other loan records of the user are good), whether special disputes or systematic errors exist or not needs to be investigated, and corresponding correction or labeling is performed.

3. Data standardization and normalization are performed on numerical data, such as age, income, enterprise income and the like, so that the numerical data has unified dimension and distribution characteristics. A common normalization method is Z-score normalization, i.e., data is converted to a standard normal distribution with a mean of 0 and a standard deviation of 1 by calculating the difference between each data point and the mean divided by the standard deviation. Thus, adverse effects on model training caused by overlarge value difference among different features can be avoided. For some data with definite value range or proportion relation, such as risk score (usually between 0 and 100) or frequency of page operation (the value represents relative proportion between 0 and 1), normalization processing can be adopted to map the data to a specific interval, so that the model can conveniently understand and compare the importance of different features.

3. User portrait tag construction:

1. Basic attribute tags are classified according to age ranges, such as 20-30 years old, 31-50 years old, and 51 years old and above. Different age tags may be further subdivided, such as young adults may be classified into 20-25 years (just walking into society), 26-30 years (business start), etc., in order to more accurately analyze loan demand and risk characteristics for users of different ages. Professional labels are classified according to industry categories (e.g., finance, internet, manufacturing, etc.), professional types (e.g., manager, technician, general staff, etc.). For example, financial industry practitioners may have a greater understanding of financial products and may be more concerned with interest rate benefits and flexible payouts in loan selection. And common staff in manufacturing industry may pay more attention to the credit line and approval speed so as to meet the demands of living consumption or purchasing vehicles in houses. Revenue level tagging-users are divided into low, medium, and high revenue groups based on local revenue level and industry average revenue. For example, in a first line city, annual revenue below 10 ten thousand yuan may be considered low revenue, 10-50 ten thousand yuan medium revenue, and more than 50 ten thousand yuan high revenue. The income level label is directly related to loan amount and repayment capability, and is an important basis for credit risk assessment and loan product recommendation.

2. Credit risk label risk score label, namely, integrating credit history information of a user, calculating a risk score of the user through a risk score model, and classifying credit grades according to the risk score, such as excellent (80-100 points), good (60-79 points), medium (40-59 points) and poor (0-39 points). The risk scoring label can intuitively reflect the overall credit condition of the user, and the higher the credit level is, the higher the loan approval passing rate is, and the lower the interest rate is. Conversely, users with low credit levels may face loan restrictions or high cost loans. The overdue risk label evaluates the overdue risk degree of the user according to the overdue times, overdue time and overdue amount of past loans of the user, and is classified into low risk (without overdue or occasional short overdue), medium risk (with a certain number of overdue records but cleared or being improved), high risk (multiple overdue and larger overdue amount and longer overdue time). The overdue risk tag may help banks to develop risk precautions in advance, such as requiring high risk users to provide additional security or increasing the pay-per-sale ratio, etc.

3. And the behavior characteristic label activity label is used for dividing the users into high-activity, medium-activity and low-activity users according to indexes such as login frequency, operation times, residence time and the like of the users on the H5 page. The highly active users may have a higher degree of interest and demand for loan products, and may be used as key marketing objects to periodically push personalized loan product information and preferential activities. The low activity user needs to further analyze the reason of inactivity, such as whether the platform experience is bad, the product is not in line with the requirement, etc., so as to take corresponding improvement measures. And the product preference label is used for determining the product preference of the user by analyzing clicking and browsing behaviors of the user on the page on different loan product types (such as consumer loan, housing loan, business loan and the like) and product characteristics (such as interest rate, repayment period length, limit size and the like). For example, a user frequently browses home loan products and pays attention to low-interest-rate and long-term product features, which may be labeled as a home loan preference user, may purposely recommend appropriate home loan products and related services for subsequent marketing.

4. Enterprise operation status tags (for enterprises owners) the enterprise scale tags divide the enterprise into small micro-enterprises, medium-sized enterprises and large enterprises according to the indexes such as the number of enterprise staff, the asset scale and the like. Enterprises of different scales have obvious differences in loan requirements, risk bearing capacity and financing channels, small micro-enterprises usually rely on external financing such as bank loans and the like, and the loan amount is relatively small. Large enterprises may have more varied financing options with more loan requirements associated with strategic expansion and capital operations. And the business stability label is used for evaluating the business stability of the enterprise according to factors such as the increase rate of the revenue, the profit stability, the market share change and the like of the enterprise and is divided into a stable increase type, a fluctuation type and a declination type. Enterprises with stable and growing operation have more advantages in loan approval, and more preferential loan conditions can be obtained. While a declining business may face higher risk assessment and more stringent loan approval procedures, banks may require the business to provide detailed business improvement plans and warranty measures.

4. User portrayal update and maintenance:

1. The periodic update mechanism sets a fixed period of time (e.g., monthly or quarterly) to update the user profile. Over time, the user's personal information, business operations, credit history, and behavioral data may all change. For example, the user's revenue may increase or decrease due to increased office pay-outs or industry changes. The revenue and profitability of an enterprise may vary with the market environment and with the adjustment of operating strategies. The credit history is updated with new loan applications and repayment records. The user's behavior on the H5 page may also vary due to platform functionality optimization, product updates, or personal demand changes. By updating the user portraits regularly, the latest state of the user can be reflected in time, and accurate and effective data used by banks in business links such as loan approval, risk assessment, product recommendation and the like are ensured.

2. In addition to periodic updating, the real-time update trigger condition should also establish a real-time update mechanism that immediately updates the user profile when certain critical events occur. For example, after a user successfully applies for and obtains a new loan, the credit history and liability change, and the related portrait labels should be updated in time. When the enterprise performs events such as significant equity change, asset reorganization or legal litigation, the information may have a significant influence on the business condition and credit risk of the enterprise, and the user portraits of the enterprise owners need to be updated immediately. In addition, some important actions of the user on the H5 page, such as submitting loan application, modifying personal information or conducting large-amount funds transaction, can also be used as real-time update triggering conditions, so that the bank can timely master the user dynamics and make corresponding business decisions.

3. The data quality monitoring and maintenance enhances the monitoring and maintenance of the data quality in the user portrait updating process. And periodically checking the integrity, accuracy and consistency of the data, and timely finding and processing the problems of data missing, abnormal value, error data and the like. Meanwhile, a data backup and recovery mechanism is established, and adverse effects on user portrait construction and application caused by data loss or damage are prevented. In addition, with the development of business and the change of data sources, the data collection and processing flow is continuously optimized, the data quality and usability are improved, and a solid guarantee is provided for the accurate construction and effective application of user portraits.

In this embodiment, the data analysis unit 132 may further include:

And the user behavior analysis unit is used for analyzing page browsing behaviors, namely tracking the browsing path of the user on the H5 page and knowing the user preference and the behavior mode. Analyzing the frequency and dwell time of the operation, analyzing the frequency and dwell time of the user in different operations (e.g. filling out forms, clicking buttons).

And the device analysis unit is used for analyzing the device type and the operating system, and identifying the device type and the operating system of the user so as to facilitate personalized service. And analyzing the geographic position, namely knowing the position information of the user through GPS data and using the position information for risk assessment and market positioning.

And the personal information analysis unit is used for analyzing the user basic information, namely analyzing personal information submitted by the user, such as age, occupation and the like, so as to evaluate loan requirements. Analyzing financial conditions, analyzing financial information submitted by the user, such as revenue, liabilities, etc., to assess repayment capabilities.

And the default prediction unit is used for predicting the future default probability of the user so as to help the bank to formulate a loan strategy.

And processing and analyzing the acquired data in real time through a server, and constructing a user portrait and a risk assessment model. The server can analyze information such as loan demands, credit conditions and the like of the user according to the behavior data of the user, and provide basis for loan approval. For example, the interests and needs of the user may be analyzed based on the user's browsing path and dwell time. And analyzing the operation habit, credit willingness and the like of the user according to the click event of the user.

The result application module 14 includes a loan approval optimizing unit 141 for optimizing a loan approval process according to the data analysis result, including adjusting the loan amount and the interest rate, and making a loan approval decision to speed up the loan approval process.

In this embodiment, the loan approval optimizing unit 141 is specifically configured to adjust the loan amount and the interest rate according to the risk score corresponding to the user, and automatically or manually make a loan approval decision according to the risk score corresponding to the user and a preset approval standard.

In the embodiment, the loan approval decision is automatically or manually made according to the risk score and the preset approval standard. The high risk loan application may be rejected or require additional review and warranty. Approval tools (such as RPA robot process automation) can be automated, and manual intervention is reduced. And (3) policy optimization, namely adjusting policies such as loan amount, interest rate, deadline and the like according to market changes and user demands.

The loan approval optimizing unit 141 is further configured to identify and eliminate redundant links and invalid operations in the approval process. And the flow sequence and the node setting are optimized, so that the approval efficiency is improved.

In this embodiment, the result application module further includes a monitoring and feedback unit, configured to monitor the approved loan in real time, and discover the potential risk in time. And (5) constructing a monitoring result and loan performance feedback model for continuous optimization of the risk assessment model. The monitoring and feedback unit is also used for establishing a risk monitoring mechanism and monitoring abnormal conditions in the loan approval process in real time. And setting an early warning threshold value, and immediately starting an emergency response mechanism once an early warning condition is triggered. And feedback and iteration, namely collecting user feedback by means of user investigation, satisfaction evaluation and the like. And analyzing the feedback data of the user to know the requirements and pain points of the user. And (3) effect evaluation, namely performing effect evaluation on the optimized approval process and strategy. Evaluation indicators include, but are not limited to, approval pass rate, approval time, user satisfaction, and the like. And (3) continuously optimizing the approval process and strategy according to the evaluation result and the user feedback.

By the adoption of the technical scheme, the loan risk can be estimated in real time, the approval efficiency and accuracy are improved, and meanwhile, the default risk is reduced.

In an alternative embodiment, the result application module 14 further includes a product optimization unit, configured to optimize, based on the data analysis result, a front end H5 page of the bank loan platform in terms of product functions, design, marketing strategies, and the like.

And the user experience improving unit is used for optimizing the user experience according to the user behavior data and improving the user satisfaction degree and the conversion rate.

And the risk control unit is used for identifying and early warning potential risks, reducing bad account rate and guaranteeing the steady operation of banking business.

In an alternative embodiment, the data embedding point design module 11 further includes:

And the custom event burying unit is used for customizing the burying point of the specific event according to the service requirement so as to capture the behavior data of the user in the specific scene.

The embedded point checking unit is used for verifying the correctness and the effectiveness of the embedded point codes and ensuring the accuracy and the completeness of data acquisition.

In an alternative embodiment, the data processing and analysis module 13 further comprises:

And the data visualization unit is used for visually displaying the analyzed data in the forms of charts, reports and the like, and is convenient for service personnel to understand and apply.

And the trend prediction unit predicts future user behaviors and market trends based on the historical data and the behavior patterns and provides data support for business decisions.

In an alternative embodiment, the information processing system based on the bank loan platform further comprises:

And the user privacy protection module is used for desensitizing the sensitive information of the user in the data acquisition, processing and storage processes, so that the safety and compliance of the user privacy are ensured.

And the permission management module is used for setting access and operation permissions of different users to system functions and guaranteeing the safety of data and the stability of the system.

In an alternative embodiment, the data embedded point design module 11 and the data acquisition module 12 are implemented through a front end framework or SDK (software development kit) to facilitate integration and maintenance of the system.

The data processing and analyzing module 13 adopts big data processing technology and machine learning algorithm to improve the efficiency and accuracy of data analysis.

In an optional implementation mode, the system can monitor the operation condition of the data buried point in real time, and comprises various links such as data acquisition, transmission, processing and storage, and the like, so that abnormal conditions can be found and processed in time, and the stability of the system and the reliability of data are ensured.

In an optional implementation manner, the information processing system based on the bank loan platform further comprises an alarm notification module, which is used for sending alarm notification to related personnel in a mail, short message or other manner when abnormal conditions are found, so as to process in time. And the log recording module is used for recording the operation log of the system and key information in the data processing process, so that the problem can be traced and checked conveniently.

The embodiment of the invention also provides a service of the front-end H5 page full-flow data embedded point based on the bank loan platform, which provides full-flow service from data embedded point design, data acquisition, processing analysis to result application, including but not limited to:

Consultation and scheme making services for data embedded point design are provided. Integration and maintenance services of the front-end SDK or framework are provided. Back-end data processing and analysis services are provided, including data cleansing, integration, visual presentation, trend prediction, and the like. Advice and policy support services providing business optimization and risk control.

According to the technical scheme, the embedded data points are embedded in the front-end H5 page, so that the real-time acquisition, transmission and analysis of behavior data of a user in the application flow of a bank loan platform are realized, accurate user portrait and risk assessment are provided for a bank, and the accuracy and efficiency of loan approval are improved.

The result application module may be used for traffic optimization:

and the approval process is optimized, namely data in the approval process are analyzed, the approval process is optimized, and the approval efficiency is improved.

Product and service optimization, namely adjusting loan products and services according to user demands and behaviors, and improving user satisfaction.

The result application module can be used for monitoring and early warning:

Monitoring abnormal behavior, namely monitoring data flow in real time, finding out abnormal behavior and triggering an early warning mechanism in time.

And monitoring performance indexes, namely monitoring key indexes of the service, such as approval speed, passing rate and the like, and ensuring the stable operation of the service.

The results application module may be used for data visualization:

And generating a real-time report, and displaying the analysis result in a chart form, so that a management layer can make decisions quickly.

And creating a management instrument panel, and displaying key business indexes and monitoring results in a centralized manner.

And the decision support is that the server processes and analyzes the behavior data of the user on the H5 page in real time, such as clicking behavior, operating frequency and the like. Decision basis, the analysis results are taken as important basis of approval decision, so that approval personnel or an automatic approval system can be helped to quickly make decision of loan approval or refusal.

Risk assessment, namely generating a risk score by analyzing user data in real time, and reflecting credit condition and loan risk of the user.

Approval criteria-risk score directly affects approval results, and users with high risk scores may get faster approval and more preferential loan conditions.

And the efficiency is improved, and real-time analysis of data can help banks identify bottlenecks in the approval process, so that the process is optimized, and the approval efficiency is improved. And the automatic approval can be realized for low-risk loan application, so that the manual intervention is reduced, and the approval speed is increased.

And the personalized service is used for customizing approval, and personalized approval and loan conditions are provided according to the risk scores and behavior data of the users. And (3) customized marketing, namely analyzing the user behavior, providing customized loan products and services for the user, and improving the marketing conversion rate.

Risk control, namely risk identification, namely a server can timely identify potential high-risk loan application and take corresponding risk control measures by analyzing data in real time. And the early warning system is established, the approval state of the loan application is monitored in real time, and the risk is found and processed in time.

And the transparency is improved, approval states and decision bases are fed back in real time, and the user experience and satisfaction are improved.

And the feedback circulation is to collect the feedback and loan performance of the user, continuously optimize the approval process and the model and improve the approval quality and the user satisfaction.

Real-time data flow, the server receives data from the H5 page in real time, such as user behavior data, personal information and the like.

And the data analysis engine is used for analyzing the real-time data (the user data collected in real time) by using the big data analysis engine to generate results such as risk scores, risk assessment and the like.

And the approval decision support system feeds back the data analysis result to the approval decision support system in real time to assist an approval person or an automatic approval system to make decisions.

And optimizing the approval process, namely optimizing the approval process according to the real-time analysis result, and improving the approval efficiency and quality.

And the user feedback circulation is to collect the feedback and loan performance of the user, continuously optimize the approval process and the model and improve the approval quality and the user satisfaction.

The technical scheme of the invention can obviously accelerate the approval process. The technical scheme of the invention has the core advantages of automation and efficiency improvement, and is specifically embodied in the following aspects:

automated approval, in which the server can automatically generate risk scores and risk assessment results by using data analyzed in real time, so that decision of loan approval or denial can be quickly made. The automatic approval process reduces manual intervention, so that the approval process is faster and more efficient.

Human errors are reduced, wherein the manual approval is possibly influenced by factors such as experience, fatigue and the like of approval personnel, and the automatic approval system is based on data and algorithms, so that the human errors are reduced, and the approval accuracy is improved.

Real-time decision support-data processing and analysis in real-time can provide real-time support for approval decisions, so that approval personnel can make decisions based on the latest data instead of obsolete information.

And the flow optimization, namely, the bottleneck in the approval flow can be identified by monitoring and analyzing the data in real time, so that the flow is optimized, and the approval efficiency is improved. For low-risk loan application, automatic approval can be realized, and approval time is reduced.

Personalized approval, namely, based on data analysis of the user, personalized approval and loan conditions can be provided, so that the approval process is quickened, and the approval quality is improved.

The content of the server for real-time processing and analysis of the acquired data not only comprises loan requirements and credit conditions, but also comprises the following aspects:

User behavior analysis, namely browsing paths and residence time, namely analyzing the browsing paths and residence time of the user on the H5 page to know the interests and the demands of the user. Click event, analyzing the operation habit and preference of the user through the action of clicking different buttons or links by the user. And (3) operating frequency, namely counting the frequency of different operations (such as filling in a form and submitting an application) of the user so as to evaluate the participation degree and the liveness of the user.

Device information, device type and operating system, identifying the device type and operating system of the user to facilitate personalizing the service. Geographic location-the knowledge of the location information of the user via GPS data for risk assessment and market location.

Personal information, basic information submitted by users, such as age, occupation, and the like, are analyzed to evaluate loan requirements. Financial conditions-analyzing financial information submitted by the user, such as revenue, liabilities, etc., to assess repayment capabilities.

Risk assessment, namely carrying out risk scoring on the user by utilizing a machine learning model so as to assess loan risk.

Breach prediction, predicting the future breach probability of the user to help the bank make loan strategy.

Business optimization, namely, approval process optimization, namely, analyzing data in the approval process, optimizing the approval process and improving the approval efficiency. Product and service optimization, namely adjusting loan products and services according to user demands and behaviors, and improving user satisfaction.

Monitoring and early warning, namely monitoring data flow in real time, finding abnormal behavior and triggering an early warning mechanism in time. And monitoring performance indexes, namely monitoring key indexes of the service, such as approval speed, passing rate and the like, and ensuring the stable operation of the service.

And the data visualization is that a real-time report is generated, and the analysis result is displayed in a chart form, so that a management layer can make decisions quickly. And creating a management instrument panel, and displaying key business indexes and monitoring results in a centralized manner.

Through real-time processing and analysis of the contents, the technical scheme of the invention can provide deep business insight for banks, optimize user experience, improve approval efficiency, reduce operation cost and effectively control risks.

The calculation mode is that firstly, risk score calculation:

Data collection, which collects multidimensional data of users, including but not limited to personal basic information (e.g., age, occupation, income, etc.), business management information (e.g., business scale, revenue, profitability, etc.), credit history (e.g., past loan records, repayment, etc.), behavioral data (e.g., operational behavior on H5 pages, residence time, etc.).

And (4) setting indexes, namely determining index weights corresponding to various data. For example, credit history may be a relatively high weight because it directly reflects the user's repayment capabilities and credit status. The enterprise management information also has a certain weight, so that the economical strength and stability of the user are reflected.

And scoring the model, namely comprehensively calculating each index by using a mathematical model. Common models are linear regression, logistic regression, decision trees, etc. Values of different indicators are translated into a specific risk score by these models.

2. Loan amount and interest rate adjustment-amount adjustment-in general, the higher the risk score, the higher the loan amount may be. Different risk scoring intervals can be set, corresponding to different credit ranges. For example, a high risk scoring user may obtain a higher upper limit of credit, while a low risk scoring user may have a relatively lower credit. At the same time, the actual demand and risk bearing capacity of the user are considered. If the user's business is well-managed and has a clear use of funds and repayment plan, the loan amount may be appropriately increased, even if the risk score is not particularly high.

Interest rate adjustment, the interest rate level is also determined based on the risk score. Users with high risk scores are at a relatively low risk and interest rates may be set lower. Users with low risk scores have higher risks and correspondingly improved interest rates. Different risk scoring intervals can be set, corresponding to different interest rates. In addition, the system can be adjusted by combining factors such as market interest rate, bank fund cost and the like. For example, at lower market rates, even users with lower risk scores may obtain relatively lower rates.

The approval accuracy is improved, namely multidimensional data evaluation, namely information of multiple aspects of users can be collected through data burial points, such as enterprise operation data (including revenue, profit, tax collection condition and the like), personal credit data (such as past loan records, repayment condition, credit rating and the like), behavior data (operation behavior, residence time and the like in an H5 page). Comprehensive these multidimensional data evaluate user's credit status and repayment ability, more accurate, objective than simply relying on one or two items of information to judge. For example, an enterprise, although having a somewhat reduced recent revenue, has recorded a good return in the past, and carefully filled out information on the H5 page, carefully looking at the relevant terms, and combining such information can more accurately assess its risk level, rather than simply refusing to loan or giving a lower credit based on the reduced revenue alone.

Dynamic real-time analysis, namely real-time data processing can reflect the current condition and change of a user in time. For example, the enterprise recently obtains a large order, or the property status of the individual is improved, and the real-time information can be captured and included in the analysis, so that the loan amount and the interest rate can be accurately adjusted. Compared with the traditional mode of periodic evaluation or relying on static data, the dynamic real-time analysis can be better suitable for market change and dynamic fluctuation of user conditions, so that the approval result is more suitable for the actual conditions of users, and the accuracy is improved.

And identifying risk signals, namely in the process of processing and analyzing data in real time, setting some risk indexes and early warning mechanisms to quickly identify potential risk signals. For example, if a user frequently queries the loan amount or interest rate in a short period of time, it may suggest that his or her funds need is urgent or that there are other potential problems. Or a certain financial index of the enterprise suddenly worsens, the system can timely find and carry out deep analysis so as to take more cautious measures such as reducing the limit or improving the interest rate during approval, thereby effectively reducing the risk of default and improving the accuracy of approval.

The approval efficiency is improved:

And an automatic decision flow is realized, namely, the loan amount and the interest rate are automatically adjusted based on the data analysis result, so that the links of manual intervention are reduced. In the traditional approval mode, the credit officer needs to spend a great deal of time to collect data, analyze and judge, and then determine the amount and the interest rate, and the process is not only inefficient, but also is easily affected by human factors, such as the experience of the credit officer, subjective judgment and the like. And the automatic decision process can be quickly calculated and adjusted according to a preset algorithm and rule after the data collection is completed, so that the approval time is greatly shortened, and the overall efficiency is improved.

The method can quickly respond to market demands and can be adjusted in real time according to market conditions and user demands. For example, when market funds are intense, banks can adjust interest rate levels in time through data analysis to attract premium customers. Or when the development potential of a certain industry is good, the system can automatically give more preferential quota and interest rate policies to enterprises in the industry, rapidly meet market demands, improve business competitiveness, improve approval efficiency and avoid customer loss caused by slow approval.

Optimizing the resource configuration, namely, the approval resource can be distributed more reasonably through analyzing a large amount of data. For applications with lower risks and more definite adjustment of the quota and the interest rate, the applications can be rapidly approved, and for applications with complex or higher risks, more manpower and time can be allocated for in-depth inspection. Therefore, the approval resources of the bank can be effectively utilized, and the overall approval efficiency is improved.

1. Service optimization and decision support:

product improvement-the number of active users may reflect the appeal and user viscosity of the product. If the number of active users is low, user behavior data can be analyzed to find out the reason of user loss, and product functions, interface design or user experience can be improved pertinently. For example, if the user is found to have high loss rate in a certain operation link, the flow of the link can be optimized, so that the link is more concise and fluent. The number of transaction users directly represents the commercial value of the product. By analyzing the behavioral characteristics, preferences and demands of the trading users, the terms, the limit range, the interest rate setting and the like of loan products can be optimized, so that the demands of the users can be better met, and the market competitiveness of the products can be improved.

Marketing strategy adjustment-the increasing trend of the number of registered users may evaluate the effectiveness of the marketing campaign. If the number of registered users grows slowly, it may be necessary to adjust marketing channels, advertising strategies, or promotional forms. For example, the input in a channel can be increased by finding that the registered user brought by the channel is high in quality according to data analysis. The change in the number of active users and the number of transacted users may help determine marketing emphasis. If the number of active users is high but the number of transacted users is low, it may be desirable to enhance the conversion marketing to the active users, such as to offer proprietary offers, provide personalized loan scheme recommendations, etc., to increase transaction conversion.

2. Risk assessment and management:

And the credit risk assessment can establish more comprehensive user portraits by combining the data of the registered user number, the active user number, the transaction user number and the like, and assist in credit assessment. For example, users who are active for a long period of time and have transaction records may be considered to be at relatively low risk, while newly registered users or users with low liveness may require more careful credit reviews. And the repayment behaviors and overdue conditions of different types of users are analyzed, so that a risk scoring model can be optimized, and the accuracy of risk identification is improved. For example, discovering that the rate of overdue for an active user of a certain class is low may give the class more preferential interest rate or a higher amount.

Fraud risk prevention, namely, potential fraud can be timely discovered by monitoring abnormal increase of registered user number, mutation of user behavior mode and the like. For example, if the number of registered users increases substantially in the short term and there is similarity or abnormal behavior in the information of these users, there may be a hint that there is a risk of batch registration fraud. The change of the number of transaction users can also be used as an early warning indicator of fraud risk. If a user suddenly makes transactions frequently or the transaction amount is abnormal, it may be necessary to further review the authenticity and rationality of their transaction.

3. Resource allocation and planning:

Human resources are arranged, namely human resources such as customer service personnel, approval personnel and the like can be reasonably arranged according to fluctuation conditions of the number of active users and the number of transaction users. During peak business hours, hands are added to ensure timely handling of user consultation and loan applications. During the business valley period, personnel training or business process optimization can be performed, and the working efficiency is improved.

And the technical resource optimization is to analyze data flow and user behavior, so that the server configuration and network bandwidth can be optimized, and the stability and response speed of the system are ensured. For example, if the user access amount is found to be concentrated in a certain period of time, the server resource can be increased in advance, so that the system is prevented from being blocked or crashed.

According to the user's requirement and the variation of using habit, the technology research and development direction is adjusted in time to develop the function and service more in line with the user's requirement. For example, if the user's demand for the mobile terminal increases, the optimization and the function expansion of the mobile terminal H5 page can be increased.

1. Data collection and analysis stage:

User behavior data is comprehensively collected, namely various behaviors of a user on an H5 page are collected through buried points, including clicking, sliding, inputting, staying time and the like. For example, the stay time of the user on different pages is recorded, and the attention of the user to each link is known. User device information, network status, etc. are collected in order to analyze external factors that may affect the user experience.

And deeply analyzing the data, namely analyzing the key node behaviors of the user in the application process. For example, looking at how many users exit halfway through the form, analysis exits may be due to complex form design, excessive filling, etc. And analyzing the page jump path to know whether the user operates according to the expected flow. If a large number of users are found to have chosen an unintended path after a certain page, this may mean that the guidance of the page is not clear enough.

2. Problem identification and optimization direction determination stage:

And identifying the problem, namely determining a specific problem affecting user experience according to the data analysis result. For example, if a user is found to stay in a certain page for too long and the jump-out rate is high, the loading speed of the page may be slow or the content may be difficult to understand.

Attention is paid to the problem that the feedback of users is more, such as complex application flow, ambiguous information prompt and the like.

And determining an optimization direction, namely, slow loading speed, optimizing a page resource loading strategy, compressing pictures, reducing unnecessary script loading and the like. For complex form design, the form content is simplified, the necessary filling items and the optional filling items are reasonably arranged, and clear filling descriptions and examples are provided.

If the user is prone to losing direction in the process, the page navigation and guidance is optimized, and explicit button labels and process instructions are used.

3. Optimization implementation and verification stage:

Implementing optimization measures:

And correspondingly adjusting and improving the H5 page according to the determined optimization direction. For example, the layout of the form is redesigned, the page loading speed is improved, and the display mode of the prompt information is optimized.

In the implementation process, the optimization measures are ensured not to introduce new problems, such as compatibility problems, abnormal functions and the like.

And verifying the optimization effect, namely burying the data again, and collecting behavior data of the user on the optimized H5 page. And comparing the data before and after the optimization, and evaluating the effect of the optimization measures. If the user residence time is shortened, the jump-out rate is reduced, the application completion rate is improved, and the like, the optimization measures are effective. If the effect is not obvious, further analysis of reasons is needed, and an optimization scheme is adjusted.

The data application scheme comprises the following steps:

Loan approval optimization, namely, application, using data analysis results (such as risk scores and repayment capability scores) to optimize a loan approval process. Optimization scheme-for high risk scoring users, loans may be automatically approved or more preferential loan conditions may be provided. For users with low risk scores, the loan may be further reviewed or rejected.

The personalized service providing application provides personalized loan products and services according to user figures and preferences. The optimization scheme is that loan products are customized for different user groups, such as low-interest rate loans are provided for students, and flexible repayment plans are provided for small enterprises.

And the risk control enhancement is applied to the application of utilizing data analysis to identify high-risk loans and taking preventive measures. Optimizing scheme, for the identified high risk loan, the guarantee requirement can be increased, the loan amount or interest rate can be adjusted, or more strict repayment monitoring can be implemented.

And marketing strategy adjustment, namely, applying to analyze the behaviors and the demands of the user and optimizing the marketing strategy. And (3) according to the user characteristics and behaviors, formulating targeted marketing activities, and improving the marketing conversion rate.

And the user experience is improved by analyzing the behavior of the user on the H5 page. The optimization scheme is used for simplifying the application flow, optimizing the page layout, providing useful guidance and help and reducing the loss of users.

The specific optimization scheme is as follows:

And the approval process is automated, namely an automatic approval system in the line is accessed, and approval decisions are automatically made based on the data analysis result.

Dynamic risk pricing, namely dynamically adjusting loan interest rate and amount according to the risk score so as to balance risk and income.

And (3) user behavior feedback loop, namely establishing a feedback mechanism, and incorporating the repayment behavior and loan performance of the user into data analysis to continuously optimize the model.

And (3) marketing by data driving, namely, utilizing a data analysis result to implement accurate marketing and improving marketing efficiency.

And optimizing the user interface, namely optimizing the design and the function of the H5 page according to the user behavior data, and improving the user satisfaction degree and the conversion rate.

And the monitoring and early warning system is used for timely finding and processing potential risks.

Continuous learning and model updating, in which the data analysis model is periodically evaluated and updated to accommodate market changes and changes in user behavior.

Through the optimization schemes, the technical scheme of the invention can more effectively utilize the data analysis result, improve the efficiency of the business process, reduce the operation cost, and improve the risk management capability and the customer service quality.

The specific risk assessment scheme comprises the following steps:

And collecting data, namely collecting behavior data and application information of the user when the user submits a loan application through an H5 page.

Feature extraction-features for risk assessment are extracted from the collected data.

Model application-these features are input into a trained risk assessment model.

Score calculation the model calculates a risk score that takes into account the user's credit history, financial status, behavioral patterns, etc.

And approval decision making, namely automatically or manually making loan approval decision making according to the risk scores and preset approval standards.

And feedback loop, wherein the approval result and the subsequent repayment behavior are fed back to the model for continuous optimization of the model.

By the method, loan risks can be effectively evaluated, approval efficiency and accuracy are improved, and meanwhile, default risks are reduced.

With the rapid development of internet technology, financial services gradually change from off-line to on-line, and on-line loan products such as bank loan platforms become an important financing tool for enterprises. However, efficient collection, analysis, and utilization of full-process data for online loan products is critical to improving product quality of service, optimizing user experience, and reducing risk. At present, the data embedding technology in the front-end H5 page has become an important means for collecting user behavior data, but the application in the financial field has certain limitations and disadvantages.

The invention provides an information processing system of a front-end H5 full-flow data embedded point based on a bank loan platform, which realizes the accurate capturing and deep analysis of the full-flow behavior of a user through the fine data embedded point design and provides powerful support for product optimization, user experience improvement and risk control. The implementation of the invention can obviously improve the service quality and user experience of online loan products, reduce risk and have wide application prospect and social value.

The method and the system aim at acquiring and integrating user behavior data and external user individual tag data to complement, and combining multidimensional information when problems occur according to product optimization, data analysis and better customized operation strategies, so that functions are correctly iterated, the product can exert the maximum effect, the marketing effect and the wind control capability are improved.

With the rapid development of the mobile internet, more and more users choose to transact financial services through the mobile phone end. As an online loan product, the bank loan platform needs to collect behavior data of a user in an application process in real time so as to perform risk assessment and construction of user portraits in order to improve user experience and approval efficiency. However, the conventional data acquisition mode often has the problems of data loss, delay and the like, and cannot meet the requirements of a bank loan platform on data timeliness and accuracy.

The invention aims to provide an information processing system based on a front-end H5 full-flow data embedded point of a bank loan platform, which realizes the accurate capture and deep analysis of the full-flow behavior of a user in a front-end H5 page of the bank loan platform through the refined data embedded point design and provides powerful support for product optimization, user experience improvement and risk control.

The aim is achieved:

(1) And (5) constructing a global user data root. And (3) completing all-channel buried point acquisition user behaviors, weChat public number self-operating entrance, H5 marketing cooperation two-dimensional code entry, industrial bank APP, unified entry system, comprehensive credit management system business process each node buried point, and integrating enterprise legal person/stakeholder/high-level external individual labels.

(2) User experience optimization. The method has the advantages that the core user use paths such as user real name authentication, credit application, repayment application and the like are optimized, the reasons of flow failure are analyzed, the user use friction is reduced, the user loss and customer complaints are reduced, and the time consumption of the user is shortened.

(3) And the marketing operation effect is improved. And accurately identifying the characteristics of the target audience and the activity of the similar lending platform by means of the external user data. By means of channel interaction data analysis of the full link, channel delivery strategies are optimized, the passenger acquisition cost is reduced, and the passenger acquisition efficiency is improved. The users are operated in a layered mode by combining the user liveness, the credit/repayment condition, the channel source and the like, and accurate recall is carried out on the users with flow failure, credit failure, good repayment condition, useful credit balance and the like. The labels of the sediment old users are complemented, and the labels comprise recent similar bank/lending platform liveness, income conditions, asset conditions, family member conditions, normal use of mobile phones and the like, and the personalized marketing awakening of multiple wave times is developed.

(4) And the wind control capability is improved. And identifying abnormal operation behaviors and frequencies of the user through buried points. The external user data labels are used for supplementing, so that wind control scenes are enriched, and normal behavior judgment of users is included, such as conventional social interaction, news browsing, automobile maintenance, live video broadcasting, billing, map navigation and the like. High risk decisions for users, such as high activity of blockchains and virtual currencies, revenue level changes, vehicle transactions, etc. And judging the risk of multi-head loan, such as high activity of similar multi-type and multi-stage loan products.

In order to achieve the above purpose, the present invention adopts the following technical scheme:

Embedding a data embedded point code in a front-end H5 page of a bank loan platform, and realizing real-time acquisition of user behaviors. The data embedded point code can collect behavior data such as clicking, inputting, sliding and the like of a user on a page, and sends the behavior data to a server for processing in real time.

And designing an index system for data embedded point acquisition, wherein the index system comprises personal information, behavior data, equipment information and the like of a user, so that the comprehensiveness and accuracy of the data are ensured.

And processing and analyzing the acquired data in real time through a server, and constructing a user portrait and a risk assessment model. The server can analyze information such as loan demands, credit conditions and the like of the user according to the behavior data of the user, and provide basis for loan approval.

And optimizing the approval process and strategy of the bank loan platform according to the data analysis result, and improving the accuracy and efficiency of approval. For example, the loan amount, interest rate, etc. may be automatically adjusted based on the risk score of the user.

The module functions can be implemented by executable programs and code development, and related product function modules and function descriptions are shown in the following table 1.

Table 1 related products function module and function introduction table

And optimizing the approval process and strategy of the bank loan platform according to the data analysis result, and improving the accuracy and efficiency of approval. For example, the loan amount, interest rate, etc. may be automatically adjusted based on the risk score of the user. By the method, personalized service and risk control of the user can be realized, and accuracy and efficiency of loan approval are improved.

The method comprises the steps of improving basic operation types, wherein analysis indexes mainly comprise active user numbers, newly-increased registered user numbers, retention rates, login times, login user numbers, login duration, delivery amount, credit amount, repayment amount and the like. The analysis dimension mainly comprises channels (cooperation institutions)/operating systems/provinces/regions/application versions/function types/user grades/product major categories/time (day/week/month) and the like. The analysis results can be visually displayed, such as the visual analysis results shown in fig. 3.

The marketing evaluation class improves analysis indexes such as popularizing a landing page PV, the number of people (UV) accessing the landing page, the jump rate of the landing page, the element/position click/browsing depth line of the landing page, the number of new channel user registration, the registration conversion rate, the new equipment retention rate and the like. The analysis dimension is mainly a popularization channel/popularization. Keyword/promotion type/active city/network type/browser type/operating system/time (day/week/month)/like lending platform high activity analysis results.

The analysis indexes mainly comprise mobile phone short message verification function distribution, faceID verification times, faceID verification user numbers, certificate uploading times, certificate uploading user numbers, scanning use times, scanning use number, average use times of a certain function, retention rate of a certain function, page retention time, jump rate, step conversion time and the like. The analysis dimension mainly comprises channels (cooperation institutions)/operating systems/provinces/regions/application versions/function types/user grades/product major categories/time (day/week/month) and the like.

Service operation class improvement: the analysis indexes mainly comprise operation site hit times, operation site hit number, personnel operation site hit times, different operation site hit numbers, personnel operation site hit times, user use paths, personal credit application flow conversion rate, enterprise credit application flow conversion rate, application credit application times, user number of application credit, credit success submission times, user number of successful credit submission, step conversion time length, personal credit application flow conversion rate, enterprise credit application flow conversion rate, credit application times, user number of application credit, credit success submission times, user number of successful credit submission times, credit success submission times (money drawing user number), credit failure submission times, credit failure submission reason distribution, application payment repayment times, application repayment times, repayment success submission times and the like. The analysis dimension mainly comprises channels (cooperation institutions)/operating systems/provinces/regions/application versions/function types/product major categories/time (day/week/month), and the like.

The innovation points of the patent are mainly embodied in the following aspects:

front end H5 full-flow data embedded point technology:

By embedding the data embedded point code in the front-end H5 page, the real-time acquisition of behavior data of the user in the application flow of the bank loan platform is realized. The technology can capture clicking, inputting, sliding and other actions of the user, and provides detailed data of user interaction for banks.

The construction of the data embedded point acquisition index system designs a set of comprehensive data embedded point acquisition index system, including personal information, behavior data, equipment information and the like of a user, and ensures the comprehensiveness and accuracy of the data. Such an index system helps to analyze user behavior and credit risk more accurately.

And (3) processing and analyzing the real-time data, namely processing and analyzing the acquired data in real time through a server, and constructing a user portrait and a risk assessment model. The real-time processing capability enables the bank to quickly respond to market changes and user demands, and improves decision-making efficiency.

And optimizing the approval process and strategy, namely optimizing the approval process and strategy of a bank loan platform according to the data analysis result, and improving the accuracy and efficiency of approval. For example, the loan amount and the interest rate can be automatically adjusted according to the risk score of the user, so that personalized service and risk control are realized.

The technology is beneficial to pushing the digital transformation of banking business, and the user experience and the service quality are improved and the competitive power of banks is enhanced by collecting and analyzing the user data in real time.

The technical scheme of the invention is not only suitable for products of a bank loan platform, but also can be widely applied to other financial business scenes and various banking business scenes, and has wide market prospect and application value.

The invention has the advantages that the user behavior data is collected in real time, and the timeliness and the accuracy of the data are ensured. And a comprehensive data index system is constructed, so that the accuracy of user portraits and risk assessment is improved. Optimizing approval process and strategy, and improving the efficiency and accuracy of loan approval. The method can be widely applied to various financial business scenes and various banking business scenes, and has wide market prospect and application value.

The information handling system of the present embodiment may be embodied in the form of functional units, where the units are ASIC (Application SPECIFIC INTEGRATED Circuit) circuits, processors and memories that execute one or more software or firmware programs, and/or other devices that provide the functionality described above.

(1) The technical scheme solves the following technical problems through the scheme of the front-end H5 full-flow data embedded point:

Flow efficiency and automation the bank realizes the rapid processing of the whole flow from client application to bank approval, which takes only 7 minutes at maximum. The improvement of the efficiency benefits from the application of the front-end data point burying technology, so that the whole flow is highly automated, and the manual intervention is reduced.

And the data driving and decision support, namely, the bank can collect and analyze the behavior data of the clients through a data embedding technology, and then the big data model is used for decision support. For example, a bank may learn the operating rate of the client device via the GPS system and analyze it in conjunction with other data to determine whether to issue a loan.

User experience optimization, namely a data embedded point technology helps a bank to optimize customer experience. The customer can complete all processes including identity authentication, information input, loan application, approval and the like through the mobile phone, and the convenience greatly improves the satisfaction of the user.

Risk control and business management the proprietary technology also helps banks make progress in risk control and business management. Through real-time monitoring and analysis of data, the bank can better evaluate and manage loan risks, and meanwhile, the speed and accuracy of business processing are improved.

In conventional financial business processes, particularly loan application and approval processes, the following methods are generally adopted for processing:

paper document processing conventional methods typically involve a large number of paper documents including loan application forms, identification, financial statements, and the like. These documents need to be filled in manually, signed, and submitted to the bank by mail or in person delivery.

Manual auditing, namely, a bank staff can manually audit the document submitted by the applicant to check the integrity and accuracy of the document. This process is typically time consuming and prone to error.

Face-to-face services-in the traditional process, customers may need to physically go to a banking branch to consult, submit application materials, and complete other related procedures.

Telephone or mail communication, in which communication between the customer and the bank is mainly performed through telephone or email, may lead to untimely information transmission or inefficient communication.

Simple data collection while some conventional systems may include basic electronic data collection functionality, these systems often lack in-depth analysis of user behavior and comprehensive monitoring of the entire application process.

Independent system operation-different business links may be handled using different systems, such as application entry systems, approval systems, post-loan management systems, etc., which may lack efficient data exchange and integration between these systems.

Compared with the traditional methods, the front-end H5 full-flow data embedded point patent technology of the technical scheme provides a more efficient, automatic and data-driven processing mode. By the technology, the bank can realize quick approval, accurate marketing, optimize user experience and improve the overall business processing capability.

(2) The defects and disadvantages of the conventional technology in the process of loan application and approval mainly include:

Inefficiency-paper documents and manual process flows are often very time consuming. Each application requires manual auditing, which can result in slow processing speeds and increased customer latency. The error rate is high, manual processing is easy to make mistakes, and whether the data is input errors or omission occurs in the document auditing process, the process delay or decision errors can be caused. The cost is high, and the maintenance of paper documents, storage space, human resources and the like all require high cost. In addition, handling errors and rework add additional costs. The user experience is poor, the customers need to communicate with the bank in person or through mails and telephones, and the process is complicated and inconvenient, so that the overall experience of the customers is affected. Data analysis and decision support is limited-traditional methods often lack deep analysis and mining of data, which limits the ability of banks to utilize data to optimize products and services, improving decision quality. Information island phenomenon is caused by the fact that effective data sharing and integration are often lacking among different systems, so that the data utilization efficiency is low. Security problems paper documents are susceptible to loss, damage, or tampering, whereas conventional data storage systems may lack adequate security measures to protect sensitive information. Lack of transparency-various links in traditional processes may not be transparent enough, and it may be difficult for clients to track application status, which may result in reduced confidence. The adaptability and the expansibility are poor, and the traditional system can be difficult to quickly adapt to new requirements along with the increase of the traffic or the market change, so that the expansibility is poor. Environmental impact-the use of paper documents in large quantities negatively impacts the environment and does not meet the requirements for sustainable development. Accordingly, to overcome these drawbacks and shortcomings, financial institutions increasingly employ digital and automated solutions, such as front-end H5 full-flow data burial techniques, to improve efficiency, reduce cost, optimize user experience, and enhance data analysis and decision support capabilities.

(3) The defects and shortcomings of the traditional technology are overcome by the following steps:

And the whole flow digitizing process is realized by transferring the whole loan application flow to an H5 platform, so that the use of paper documents is reduced, and the processing speed is improved.

And (3) embedding and monitoring the data in real time, namely, embedding the data on the H5 page, and collecting user behavior data in real time to provide instant business insight for banks, so that the process is optimized and the user experience is improved.

And in the automatic approval process, the collected data is combined with a machine learning algorithm to realize the automation of loan approval, so that the manual intervention is reduced, the error rate is reduced, and the approval efficiency is improved.

The H5 technology supports the user-friendly interface design, so that the loan application process is more visual and convenient, and the operation experience of a user is improved.

And the data integration and analysis are that the information island is broken through by integrating the data of different systems, so that the bank can perform more comprehensive data analysis and risk assessment.

The security is enhanced, namely, the transmission and storage security of the user data on the H5 platform are ensured by adopting an advanced encryption technology and a security protocol.

The cost is saved, the cost of processing, storing and manually checking the paper documents is reduced, and the operation cost is further reduced through an automatic process.

The environment-friendly paper document reduces the dependence on the paper document, is beneficial to environmental protection and accords with the concept of sustainable development.

Flexibility and expansibility the flexibility of the H5 platform enables the bank to quickly adjust and expand services according to market demands, and improves service adaptability.

(4) Specifically, the following are several key points for solving the drawbacks and shortcomings of the technical scheme of the present invention:

And the rapid approval is realized by the immediate data collection and automatic approval of the H5 platform, so that the loan approval time is greatly shortened from the traditional days to 7 minutes at maximum.

And the decision support is to carry out loan decision based on user behavior data by utilizing big data analysis and machine learning, so that the accuracy and the efficiency of the decision are improved.

The user experience is that the user can finish the whole loan flow on the mobile phone without going to a bank, so that the operation steps are simplified, and the satisfaction is improved.

And risk control, namely, through real-time monitoring and data analysis, the bank can better manage loan risk and ensure stable business.

(5) The specific analysis mode for carrying out real-time processing and analysis on the acquired data through the data processing and analysis module on the back-end server in the technical scheme of the invention may comprise the following aspects:

user behavior analysis 1) click stream analysis, namely tracking the click behavior of a user, and analyzing the interaction mode of the user on an H5 page, such as which buttons are clicked and which pages are accessed. 2) Path analysis, namely analyzing paths from entering application to finishing loan application by users, and finding out common user behavior paths and potential loss points.

Data mining, 1) feature extraction, namely extracting key features such as personal information, equipment information, behavior patterns and the like of a user from the acquired data. 2) Pattern recognition-using a machine learning algorithm to identify patterns in user actions, such as which actions are related to loan approval passing rates.

Machine learning model 1) risk scoring model, namely, establishing a risk scoring model based on user data by utilizing a machine learning technology such as random forest, gradient elevator and the like for automatic approval process. 2) And the risk prediction model predicts the loan default risk and helps the bank determine whether to approve the loan application.

Real-time monitoring and early warning, 1) abnormal detection, namely real-time monitoring a data stream, finding abnormal behaviors by using an abnormal detection algorithm, and triggering an early warning mechanism in time. 2) And the performance index monitoring is used for monitoring the performance index of the loan approval process, such as approval speed, passing rate and the like.

Statistical analysis 1) conversion analysis, namely calculating the conversion rate of key links from visit to application, from application to approval and the like. 2) A/B test analysis, namely testing different page layouts or flow designs, and analyzing which mode is more effective.

And 2) grouping the users, namely dividing the users into different groups according to the behavior characteristics, the credit conditions and the like of the users, so as to realize accurate marketing. 2) And analyzing market trend and change of user demand, and providing data support for product iteration and market strategy.

And 2) data visualization, namely, generating a real-time report, and displaying the analysis result in a chart form so as to facilitate a management layer to quickly make decisions. 3) And creating a management instrument panel, and displaying key business indexes and monitoring results in a centralized manner.

The analysis modes are combined, so that deep business insight can be provided for operation of banks, user experience is optimized, approval efficiency is improved, operation cost is reduced, and risks are effectively controlled.

(6) The process of constructing the user representation and risk assessment model is generally as follows:

Constructing user portraits 1) data collection, namely collecting basic information such as age, gender, occupation and the like of users. Transaction behavior data, browsing behavior, click data and the like of the user are acquired. 2) Feature engineering, extracting useful features from the collected data, such as consumption habits, repayment capabilities, social activities, and the like. And cleaning, converting and normalizing the characteristics. 3) User clustering, namely clustering the users by using a clustering algorithm (such as K-means) to find out groups with similar characteristics. Tags are defined for each group, such as "high-income group", "student group", and the like. 4) And constructing detailed portraits of each user according to the characteristics and the grouping result. The user representation may include multiple dimensions, such as demographics, consumption features, credit features, and the like.

The risk assessment model is constructed 1) data preparation, namely historical loan data is collected, and the historical loan data comprises loan amount, repayment condition, overdue records and the like. In conjunction with the user portrayal data, a complete data set is prepared for modeling. 2) Model selection, namely selecting a proper machine learning algorithm such as logistic regression, decision trees, random forests and the like. 3) And selecting the features with the greatest influence on risk assessment through feature importance analysis. 4) Model training, training the model by using the training data set, and adjusting model parameters to achieve the best performance. The accuracy and generalization ability of the model are evaluated by methods such as cross-validation. 5) Model verification, namely testing the model by using a verification data set, and ensuring the effectiveness of the model in practical application.

Subsequent use 1) use of the user representation, personalization services, providing personalized loan products and services based on the user representation. Marketing strategy, namely, aiming at the characteristics of different user groups, setting accurate marketing strategy. And (3) optimizing the user experience, namely improving the design and the function of the H5 page according to the user portrait, and improving the user experience. 2) And the risk assessment model is used for automatically approving, namely when a user submits a loan application, the risk assessment model is automatically used for approving, so that the approval efficiency is improved. Risk control, namely identifying a high risk loan application, and taking measures to reduce potential default risks. Decision support, namely providing data support for bank management personnel and helping to make more intelligent loan decisions. By the method, the bank can manage loan business more effectively, improve service quality and reduce operation cost and risk. The server can analyze information such as loan requirements, credit conditions and the like of the user according to the behavior data of the user.

(7) The scoring type is that in the technical scheme of the invention, the scoring may comprise the following types, namely the scoring of the loan requirement reflects the requirement degree of the user on the loan and may be based on the behavior mode, the application information, the historical loan record and other factors of the user. Credit status score this score evaluates the credit risk of the user, typically based on the user's payment record, credit history, financial status, etc.

In addition to the above scores, there may be several scores, repayment capability scores, which evaluate the ability of the user to repay the loan, possibly taking into account the user's income level, liability status, etc. Fraud risk scoring, the assessment of the authenticity of information submitted by the user for the detection of potential fraud. And comprehensively scoring, namely combining the plurality of scoring dimensions to provide a comprehensive assessment result for the user.

These scoring results may be used to automatically approve, by automatically determining whether to approve the loan application based on the scoring results. Risk pricing, namely adjusting loan interest rate or amount according to the risk score. And personalized service, providing customized loan products according to the loan requirement scores.

(8) The influence of the mode for providing basis for loan approval on the approval process is mainly represented by the following aspects:

1) And the approval efficiency is improved, namely the automatic processing is realized, and the credit condition and repayment capability of the loan user can be automatically evaluated by collecting and analyzing the behavior data of the user in real time. The automatic scoring model can rapidly give approval suggestions, and reduces the time of manual approval, so that the approval efficiency is remarkably improved.

2) The approval accuracy is enhanced, namely, a data-driven decision is made by utilizing big data and a machine learning technology, and the approval process is not simply dependent on manual judgment but is analyzed based on a large amount of data and a complex algorithm. This data-driven decision-making approach enables more accurate identification of potential premium customers and high risk customers.

3) The fine management of risk control, namely the data embedded technology enables a bank to collect more dimensional information such as browsing behaviors, operation habits and the like of a user, and the information is helpful for finer risk assessment. By analyzing this data, the bank can better identify and prevent fraud, reducing the risk of loan violations.

4) The approval process is transparent, namely the process monitoring, namely the patent technology allows a bank to monitor each link of the approval process in real time, so that the transparency and fairness of the process are ensured. The management layer can check the approval state and the decision basis at any time, and is convenient for supervising and adjusting the approval strategy.

5) The method comprises the specific mode of influence, namely, after a user submits an application, the system immediately performs preliminary screening according to preset rules and models, and applications which do not meet the conditions are rapidly eliminated. Risk score-the risk score of the user directly affects the approval result. A high risk scoring user may obtain faster approval and more preferential loan conditions. And analyzing the behavior patterns of the user on the H5 page, such as repeatedly checking loan conditions, hesitating, and the like, wherein the behavior characteristics can be used as reference factors for approval. And the approval model can dynamically adjust approval standards according to market conditions and bank strategies, so that the flexibility and adaptability of the approval process are ensured. Instant feedback the system may provide approval feedback to the user on-the-fly and may provide improvement advice or alternatives to the rejected application.

(9) And the loan amount is adjusted in a data collection and analysis mode, namely collecting basic information, financial conditions, historical credit records, behavior data and the like of the user. These data are analyzed to assess the repayment capabilities, credit risk, etc. of the user. Risk scoring model-training a risk scoring model based on user data using a machine learning algorithm. The risk scoring model outputs a score reflecting the credit rating of the user. And setting a credit rule, namely determining the loan credit limit ranges corresponding to different credit levels according to the risk scores and preset rules. Rules may include credit score versus credit limit mapping tables, risk tolerance, and the like. Dynamic adjustment, namely dynamically adjusting the limit rule according to market conditions, bank risk preference and policy guidance. For a particular user group or market activity, a temporary credit adjustment policy may be implemented. User feedback and adjustment, namely adjusting the loan amount of an individual user according to the feedback and the loan performance of the user.

(10) And the interest rate adjustment mode is that a risk pricing model is established, and credit risk, market interest rate, bank cost and other factors of the user are taken into consideration. The model calculates the risk-based loan interest rate. And setting the interest rate rule, namely setting the interest rate rule and corresponding the risk score to the interest rate level. Rules may include interest rate intervals for different credit levels, fixed addition or float ratios, etc. And (3) market interest rate linkage, namely, taking market interest rate change, such as reference interest rate change, into consideration, and correspondingly adjusting loan interest rate. Personalized pricing, namely providing personalized interest rate pricing according to factors such as risk scores of users, loan histories, relations with banks and the like. Promotion and offer-promotion interest rate or offer conditions are provided at a particular time period or for a particular group of users.

And the specific adjustment flow is application submission, namely a user submits a loan application through an H5 page. Data assessment-the server collects and analyzes user data, including behavioral data, risk scores, and the like. And determining the credit limit and the interest rate, wherein the system automatically determines the credit limit and the interest rate according to the risk scores and the preset rules. Approval feedback, namely feeding approval results (including the amount and the interest rate) back to the user. The user confirms that the user accepts or rejects the offered loan terms.

By the adoption of the technical scheme, dynamic and personalized adjustment of the loan amount and the interest rate can be realized, so that the method is suitable for risk conditions and market environments of different users, and simultaneously, the loan business efficiency and risk management of banks are optimized.

It should be noted that what is not described in detail in the present specification belongs to the known technology of those skilled in the art.

Fig. 4 is a schematic structural diagram of an information processing method based on a bank loan platform according to an embodiment of the invention.

As shown in FIG. 4, the embodiment of the invention provides an information processing method based on a bank loan platform, which comprises S101, determining a data embedded point scheme according to the business flow and the user behavior characteristics of a front-end H5 page of the bank loan platform. The business process includes a loan application process. S102, acquiring user data in real time according to a data embedding scheme, and transmitting the user data to a back-end server. The user data includes behavior data and application information of the user. The behavior data includes a browsing path, a stay time, and a click event of the user. The application information comprises application time, loan amount, loan period, repayment mode, personal information, such as the name, age and sex of the user, enterprise operation information, enterprise scale and operation information, credit history, historical loan record and historical repayment record. S103, receiving and processing analysis user data to determine data analysis results. And S104, optimizing the loan approval process according to the data analysis result, wherein the method comprises the steps of adjusting the loan amount and the interest rate, and making a loan approval decision so as to accelerate the loan approval process. Reference is made in detail to specific implementation manners of the above system, and details are not repeated here.

Referring to fig. 5, fig. 5 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present invention, and as shown in fig. 5, the computer device includes one or more processors 10, a memory 20, and interfaces for connecting components, including a high-speed interface and a low-speed interface. One processor 10 is illustrated in fig. 5. The processor 10 may be a central processor, a network processor, or a combination thereof. The memory 20 stores instructions executable by the at least one processor 10 to cause the at least one processor 10 to perform methods that implement the embodiments shown above. The computer device also includes a communication interface 30 for the computer device to communicate with other devices or communication networks.

The embodiments of the present invention also provide a computer readable storage medium, and the method according to the embodiments of the present invention described above may be implemented in hardware, firmware, or as a computer code which may be recorded on a storage medium, or as original stored in a remote storage medium or a non-transitory machine readable storage medium downloaded through a network and to be stored in a local storage medium, so that the method described herein may be stored on such software process on a storage medium using a general purpose computer, a special purpose processor, or programmable or special purpose hardware. The method shown in the above embodiment is implemented.

Portions of the present invention may be implemented as a computer program product, such as computer program instructions, which when executed by a computer, may invoke or provide methods and/or aspects in accordance with the present invention by way of operation of the computer. Those skilled in the art will appreciate that the existence of computer program instructions in a computer-readable medium includes, but is not limited to, source files, executable files, installation package files, and the like, and accordingly, the manner in which computer program instructions are executed by a computer includes, but is not limited to, the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled programs, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed programs. Herein, a computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by a computer.

Although embodiments of the present invention have been described in connection with the accompanying drawings, various modifications and variations may be made by those skilled in the art without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope of the invention as defined by the appended claims.

Claims (10)

1. An information processing system based on a bank loan platform, comprising:

The data embedded point design module is used for determining a data embedded point scheme according to the business process and the user behavior characteristics of the front end H5 page of the bank loan platform, wherein the business process comprises a loan application process;

the data acquisition module is configured in the front-end H5 page and is used for acquiring user data in real time according to the data embedding scheme determined by the data embedding design module and transmitting the user data to the rear-end server;

The user data comprises user behavior data and application information, wherein the behavior data comprises a browsing path, a stay time and a clicking event of the user, the application information comprises application time, loan amount, loan period, repayment mode, personal information, enterprise operation information, enterprise scale and operation information, credit history, historical loan records and historical repayment records;

the data processing and analyzing module is configured at the back-end server and is used for receiving and processing and analyzing the user data to determine a data analysis result;

The result application module comprises a loan approval optimizing unit, and is used for optimizing a loan approval process according to the data analysis result, wherein the loan approval optimizing unit is used for adjusting the loan amount and the interest rate and making a loan approval decision so as to accelerate the loan approval process.

2. The system of claim 1, wherein the data burial point design module comprises:

The embedded point demand determining unit is used for determining the embedded point demand according to the business flow and the user behavior characteristics of the front end H5 page of the bank loan platform, wherein the embedded point demand comprises key nodes and data fields corresponding to the key nodes;

the embedded point scheme design unit is used for determining embedded point positions according to the key nodes in the embedded point requirements, determining data types according to the data fields corresponding to the key nodes, and designing corresponding acquisition frequencies.

3. The system of claim 1, wherein the data acquisition module comprises:

The data acquisition unit is used for acquiring user data in real time according to the data embedding point scheme determined by the data embedding point design module;

the data encryption unit is used for carrying out encryption processing on the user data to obtain encrypted user data;

the data compression unit is used for compressing the encrypted user data to obtain compressed user data;

And the data transmission unit is used for transmitting the compressed user data to the back-end server.

4. The system of claim 1, wherein the data processing and analysis module comprises:

The data processing unit is used for receiving the user data and carrying out data processing on the user data to obtain the user data after the data processing, wherein the data processing comprises preprocessing and feature extraction;

and the data analysis unit is used for analyzing the user data after the data processing to obtain the data analysis result.

5. The system of claim 4, wherein the data analysis unit comprises:

the model building unit is used for building a risk assessment model;

the model application unit is used for inputting the user data after data processing into the risk assessment model as input data to obtain risk scores corresponding to the users, and the risk scores are used for representing credit risks and loan risks of the users.

6. The system according to claim 5, wherein the loan approval optimizing unit is specifically configured to adjust a loan amount and an interest rate according to a risk score corresponding to a user, and automatically or manually make a loan approval decision according to the risk score corresponding to the user and a preset approval standard.

7. The system of claim 5, wherein the data analysis unit further comprises:

And the user portrait construction unit is used for analyzing the user data after the data processing to construct a user portrait.

8. An information processing method based on a bank loan platform is characterized by comprising the following steps:

Determining a data embedded point scheme according to the business process of the front end H5 page of the bank loan platform and the behavior characteristics of a user, wherein the business process comprises a loan application process;

the method comprises the steps of collecting user data in real time according to a data burial point scheme and sending the user data to a back-end server, wherein the user data comprises behavior data of a user and application information, the behavior data comprises a browsing path, stay time and clicking events of the user, the application information comprises application time, loan amount, loan deadline, repayment mode and personal information, name, age and sex of the user, enterprise operation information comprises enterprise scale and operation information, and credit history comprises a history loan record and a history repayment record;

receiving and processing and analyzing the user data to determine a data analysis result;

optimizing the loan approval process according to the data analysis result comprises adjusting the loan amount and the interest rate, and making a loan approval decision so as to accelerate the loan approval process.

9. A computer device, comprising:

the information processing method based on the bank loan platform comprises a memory and a processor, wherein the memory and the processor are in communication connection, the memory stores computer instructions, and the processor executes the computer instructions, so that the information processing method based on the bank loan platform according to claim 8 is executed.

10. A computer-readable storage medium having stored thereon computer instructions for causing a computer to execute the bank loan platform-based information processing method of claim 8.