CN111754253A - User authentication method, device, computer equipment and storage medium - Google Patents

Abstract

The application provides an authentication method, an authentication device, computer equipment and a storage medium for a user, wherein the method comprises the following steps: determining a target feature class related to a target service based on first preset sample data; receiving an authentication request, wherein the authentication request comprises an identifier of a user to be authenticated; acquiring related data of the user to be authenticated according to the identification of the user to be authenticated, wherein the related data comprises second feature data of each target feature class corresponding to the user to be authenticated; authenticating the user to be authenticated based on the related data of the user to be authenticated; and if the authentication is passed, pushing the content related to the target service to the user to be authenticated. The target characteristic class relevant to the target service is determined from the plurality of characteristic classes of the user group, the characteristics of the target service are expanded, and the potential users of the target service are authenticated and pushed further based on the expanded target characteristic class, so that the mining amount and the accuracy of the potential users are improved, the accuracy of pushing service information can be effectively improved, and the marketing effect is improved.

Description

User authentication method, device, computer equipment and storage medium

technical field

The present application relates to the field of Internet technologies, and in particular, to a user authentication method, device, computer equipment and storage medium.

Background technique

With the continuous development of e-commerce business, the expansion of new categories has become a trend of various e-commerce platforms. For example, from shopping categories to rental service categories, rental business is a new business direction of e-commerce platforms, and the development of new businesses requires Mining demand users for precise marketing and improving the conversion rate of potential users.

In the prior art, based on the user's online rental behavior data, a prediction model is used to predict the potential conversion rate of the user group to achieve marketing purposes. However, for new businesses, the number of training samples that can be used to train the prediction model is limited, resulting in a low mining volume of potential users and poor marketing effects for new businesses.

SUMMARY OF THE INVENTION

The present application provides a user authentication method, device, computer equipment and storage medium, so as to solve the defects of the prior art, such as the small amount of mining potential users and the inaccurate determination.

A first aspect of the present application provides a user authentication method, including:

determining a target feature class related to the target service based on the first preset sample data, where the first preset sample data includes the first feature data of the user's historical behavior;

receiving an authentication request, where the authentication request includes the identification of the user to be authenticated;

Obtain relevant data of the user to be authenticated according to the identification of the user to be authenticated, and the relevant data includes the second feature data of each of the target feature classes corresponding to the user to be authenticated;

Authenticating the user to be authenticated based on the relevant data of the user to be authenticated;

If the authentication is passed, the content related to the target service is pushed to the user to be authenticated.

Optionally, determining the target feature class related to the target service based on the first preset sample data, including:

According to the first feature data and the preset random forest RF model, determine the target feature importance score of each feature class in the first feature data in the RF model;

According to the target feature importance score of each feature class in the RF model, the target feature class related to the target service is determined.

Optionally, according to the first feature data, determine the target feature importance score of each feature class in the first feature data in the RF model, including:

obtaining first samples corresponding to at least two decision trees according to the first feature data;

For each decision tree, based on its corresponding first sample, determine the first feature importance score of m random feature classes among the M feature classes in the decision tree, where M is the feature included in the first feature data The total number of classes, m is an integer less than M;

For each feature class, based on the first feature importance score of the feature class in each decision tree, determine the second feature importance score of the feature class, and the second feature importance score is the feature class in each decision tree. The sum of the first feature importance scores in the tree;

The second feature importance score of each feature class is normalized to obtain the target feature importance score of each feature class.

Optionally, for each decision tree, based on its corresponding first sample, determine the first feature importance score of m random feature classes in the decision tree among the M feature classes, including:

For each node in the decision tree, calculate the Gini index of each feature class of the m feature classes at the node;

For each feature class in the m feature classes, based on the Gini index of each node of the feature class in the decision tree, calculate and obtain the variation of the Gini index of the feature class at each node;

Based on the variation of the Gini index of each of the m feature classes at each node, the first feature importance score of the m feature classes in the decision tree is calculated and obtained.

Optionally, according to the target feature importance score of each feature class in the RF model, determine the target feature class related to the target service, including:

According to the target feature importance score, a feature class with a score greater than a preset threshold is obtained as the target feature class.

Optionally, the target business is a rental business, and the first feature data includes rental behavior feature data and non-rental behavior feature data mined based on the rental behavior feature data.

Optionally, the authentication of the user to be authenticated based on the relevant data of the user to be authenticated includes:

Based on the relevant data of the user to be authenticated and the pre-trained classification model, the user to be authenticated is authenticated.

Optionally, the classification model is an Xgboost model;

Before authenticating the user to be authenticated based on the relevant data of the user to be authenticated and the pre-trained classification model, the method further includes:

Get training sample data;

The pre-established Xgboost network is trained by using the training sample data to obtain the Xgboost model.

A second aspect of the present application provides an authentication device for a user, including:

a determination module, configured to determine a target feature class related to the target service based on first preset sample data, where the first preset sample data includes first feature data of user historical behavior;

a receiving module, configured to receive an authentication request, the authentication request including the identification of the user to be authenticated;

an acquisition module, configured to acquire the relevant data of the user to be authenticated according to the identification of the user to be authenticated, the relevant data including the second feature data of each of the target feature classes corresponding to the user to be authenticated;

An authentication module, configured to authenticate the user to be authenticated based on the relevant data of the user to be authenticated;

A push module, configured to push the content related to the target service to the user to be authenticated if the authentication is passed.

Optionally, the determining module is specifically used for:

According to the first feature data and the preset random forest RF model, determine the target feature importance score of each feature class in the first feature data in the RF model;

According to the target feature importance score of each feature class in the RF model, the target feature class related to the target service is determined.

Optionally, the determining module is specifically used for:

obtaining first samples corresponding to at least two decision trees according to the first feature data;

For each decision tree, based on its corresponding first sample, determine the first feature importance score of m random feature classes among the M feature classes in the decision tree, where M is the feature included in the first feature data The total number of classes, m is an integer less than M;

For each feature class, based on the first feature importance score of the feature class in each decision tree, determine the second feature importance score of the feature class, and the second feature importance score is the feature class in each decision tree. The sum of the first feature importance scores in the tree;

The second feature importance score of each feature class is normalized to obtain the target feature importance score of each feature class.

Optionally, the determining module is specifically used for:

For each node in the decision tree, calculate the Gini index of each feature class of the m feature classes at the node;

For each feature class in the m feature classes, based on the Gini index of each node of the feature class in the decision tree, calculate and obtain the variation of the Gini index of the feature class at each node;

Based on the variation of the Gini index of each of the m feature classes at each node, the first feature importance score of the m feature classes in the decision tree is calculated and obtained.

Optionally, the determining module is specifically used for:

According to the target feature importance score, a feature class with a score greater than a preset threshold is obtained as the target feature class.

Optionally, the target business is a rental business, and the first feature data includes rental behavior feature data and non-rental behavior feature data mined based on the rental behavior feature data.

Optionally, the authentication module is specifically used for:

Based on the relevant data of the user to be authenticated and the pre-trained classification model, the user to be authenticated is authenticated.

Optionally, the classification model is an Xgboost model; the acquisition module is also used for:

Get training sample data;

The pre-established Xgboost network is trained by using the training sample data to obtain the Xgboost model.

A third aspect of the present application provides a computer device, comprising: at least one processor and a memory;

The memory stores a computer program; the at least one processor executes the computer program stored in the memory to implement the method provided by the first aspect.

A fourth aspect of the present application provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed, the method provided in the first aspect is implemented.

The user authentication method, device, computer equipment and storage medium provided by the present application, by adopting a preset model, determine the target feature class related to the target service from the numerous feature classes of the user group, expand the feature of the target service, and further based on The expanded target feature class obtains the relevant data of the user to be authenticated according to the identity of the user to be authenticated, authenticates the user to be authenticated based on the relevant data of the user to be authenticated, and sends the authentication to the authentication user. Users to be authenticated who have passed the authorization can push content related to the target business, which improves the mining volume and accuracy of potential users, and can effectively improve the accuracy of business information push, thereby improving the marketing effect.

Description of drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the accompanying drawings used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are some embodiments of the present application, and for those of ordinary skill in the art, other drawings can also be obtained from these drawings without any creative effort.

1 is a schematic flowchart of a user authentication method provided by an embodiment of the present application;

2 is a schematic flowchart of a user authentication method provided by another embodiment of the present application;

3 is a schematic diagram of an overall authentication process provided by an embodiment of the present application;

4 is a schematic structural diagram of a user authentication device provided by an embodiment of the present application;

FIG. 5 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Specific embodiments of the present application have been shown by the above-mentioned drawings, and will be described in more detail hereinafter. These drawings and written descriptions are not intended to limit the scope of the disclosed concepts in any way, but to illustrate the concepts of the present application to those skilled in the art by referring to specific embodiments.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments It is a part of the embodiments of the present application, but not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

First, the terms involved in this application are explained:

Random Forest: Random Forest, RF, refers to a classifier that uses multiple decision trees to train and predict samples.

Xgboost: Xgboost is an efficient implementation of the GB (Gradient boosting) algorithm. The base learner in xgboost can be a CART (gbtree) or a linear classifier (gblinear).

The user authentication method provided by the embodiment of the present application is suitable for the scenario in which the e-commerce platform mines potential users in a new business or an original business, such as a rental business, combined with the direct behavioral feature data of the user group in the target business and based on direct Other behavioral feature data mined from behavioral data is used to determine important feature classes related to the target business. For example, according to the user's browsing frequency, days, attention, search, etc., to find out the useful characteristics of potential users, such as which other categories of pages have been browsed by users who generate rental behavior, for example, a certain percentage of users have also browsed In terms of maternal and child and home improvement, the behaviors of these groups in the category of browsing, searching, following, and placing orders can be used as potentially useful features. Another example is based on the location information (such as longitude and latitude coordinates) of the user when browsing the user's address, and according to the user's work and rest time, combined with the user's age, occupation, and whether it is a college student or white-collar work address change, etc., this part of the user group can also be included. potential crowd, etc.

In addition, the terms "first", "second", etc. are used for descriptive purposes only, and should not be construed as indicating or implying relative importance or implying the number of indicated technical features. In the description of the following embodiments, the meaning of "plurality" is two or more, unless otherwise expressly and specifically defined.

The following specific embodiments may be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. Embodiments of the present invention will be described below with reference to the accompanying drawings.

Example 1

This embodiment provides a user authentication method for authenticating potential users of a target service. The execution subject of this embodiment is the user's authentication device, and the device may be set in a computer device, such as a server, a desktop computer, a notebook computer, etc., or may be set in an e-commerce platform.

As shown in FIG. 1 , a schematic flowchart of a user authentication method provided in this embodiment, the method includes:

Step 101: Determine a target feature class related to the target service based on first preset sample data, where the first preset sample data includes first feature data of historical user behavior.

Specifically, the first preset sample data may include direct feature data of the target service and indirect feature data based on direct feature data mining. The target business can be a new category of business launched by the e-commerce platform, such as the rental business, or it can be an original business.

Taking the rental business as an example, the first preset sample data may include behavioral characteristic data such as the frequency, days, attention, search, etc. of the user group browsing on the rental category, as well as the browsing, Search, follow, order behavior characteristic data and user’s location information when browsing, work and rest time, age, occupation, and based on these analysis, whether the address of college students or white-collar jobs has changed, the distance after the address change, etc. Include these users in the potential population as well.

Exemplarily, according to the user's historical behavior characteristic data on the rental category, the users who have completed the conversion of rental users in the rental category are counted, and the behavior characteristics data of these users in other categories on the e-commerce platform, such as browsing, searching, following, are obtained. , behavioral characteristics data of other products placed, and obtain useful characteristics of potential users through statistical analysis. These features can be incorporated into the useful features of potential users. For another example, the user’s address can be reversely checked through the user’s location information when browsing. According to the user’s work and rest time, combined with the user’s age and occupation, it can be determined whether the address is changed for college students or white-collar workers, and the address The changed distance, etc. For example, in the graduation season, college students need to rent a house after graduation. According to the location information of the user when browsing, it can be reversed that the user's address is the school, and combined with the user's age, it can be judged that they may graduate this year. These features are also included in the useful features of potential users, and in a similar way, a group of useful features of potential users are mined based on existing data and experience. That is, the first preset sample data includes various collected useful feature data of potential users (ie, the first feature data). The specific collection process of the useful features of potential users may be obtained by a person based on experience, or may be obtained by statistical analysis of existing data, and the specific manner is not limited.

After the useful features of these potential users are collected, the historical user behavior data related to the useful features of these potential users can be obtained from the historical user data of the e-commerce platform to form the first preset sample data, which is used to obtain useful features from these potential users. An important feature class (ie, target feature class) related to the target business is determined in the features, and the feature class represents a class of features, for example, user age is a class of features, and user gender is a class of features. The feature data is user data representing feature values including each feature class, such as "Zhang San-male-23-year-old-student".

Exemplarily, a total of 800 feature classes are obtained, including the behavior characteristic data of renting and non-renting behavior. In the follow-up, it is necessary to determine a batch of target feature classes related to the rental business from these 800 feature classes. For example, it is determined that 200 characteristic categories are relatively related to the rental business and so on.

Optionally, a preset random forest RF model may be used to determine a target feature class related to the target service from feature classes included in the first feature data. Other algorithm models may also be used to determine the target feature class, and the specific manner is not limited in this embodiment.

Step 102: Receive an authentication request, where the authentication request includes the identifier of the user to be authenticated.

Specifically, when the merchant needs to authenticate which users are potential users, an authentication request may be sent to the user's authentication device through the terminal, and the authentication request may include the identifier of the user to be authenticated. The identification of the user to be authenticated may be the account registered by the user on the e-commerce platform, such as the user ID assigned to the user, or other forms of identification, as long as the user can be uniquely identified in the e-commerce platform. The authentication request may include the identifiers of one or more users to be authenticated. Usually it can be a batch of users to be authenticated. For example, the corresponding management personnel of the merchant collect all or part of the user IDs registered on the e-commerce platform, and need to authenticate the potential users of the target business from these, then the management personnel can use the terminal interface. An authentication request is triggered, and these user IDs are carried in the authentication request as the identifiers of the users to be authenticated.

Step 103: Acquire relevant data of the user to be authenticated according to the identifier of the user to be authenticated, where the relevant data includes second feature data of each target feature class corresponding to the user to be authenticated.

Specifically, after receiving the authentication request, the identification of the user to be authenticated can be obtained from the authentication request, and the relevant data of the user to be authenticated can be further obtained according to the identification of the user to be authenticated, and the relevant data includes the user to be authenticated. Corresponding second feature data of each target feature class. Relevant data can be extracted from user historical data of related categories. For example, from the user's historical data in October, the second characteristic data is extracted to form relevant data. Used to authenticate target potential users in November, etc. The specific relevant data can be obtained according to actual needs, which is not limited in this embodiment.

In step 104, the user to be authenticated is authenticated based on the relevant data of the user to be authenticated.

Specifically, after obtaining the relevant data of the user to be authenticated, the user to be authenticated can be authenticated based on the relevant data of the user to be authenticated, and whether the user to be authenticated is a potential user of the target service can be authenticated. Specifically, a pre-trained classification model may be used to authenticate the user to be authenticated.

Exemplarily, the target feature class includes 100 types, that is, each second feature data is 100-dimensional, and the second feature data includes 10,000 pieces. Input the second feature data into the pre-trained classification model, and authenticate a batch of targets. Potential users, for example, out of 10,000 pieces of data authenticated, 2,000 users are targeted potential users. That is, the authentication result for these 2000 users is that the authentication is passed.

Step 105, if the authentication is passed, push the content related to the target service to the user to be authenticated.

Specifically, after the target potential user is authenticated, the content related to the target service can be pushed to the user to be authenticated who has passed the authentication. , or push relevant information to the user's mobile phone, etc. The specific push mode can be set according to actual needs, which is not limited in this embodiment.

In the user authentication method provided by this embodiment, the target feature class related to the target service is determined from the numerous feature classes of the user group by adopting a preset model, the feature of the target service is expanded, and further based on the expanded target feature class, Obtain the relevant data of the user to be authenticated according to the identification of the user to be authenticated, authenticate the user to be authenticated based on the relevant data of the user to be authenticated, and push the content related to the target service to the user to be authenticated that has passed the authentication, improving the It can effectively improve the accuracy of business information push, thereby improving the marketing effect.

Embodiment 2

This embodiment further supplements the description of the method provided in the first embodiment.

As shown in FIG. 2 , it is a schematic flowchart of a user authentication method provided in this embodiment.

As an implementable manner, on the basis of the foregoing Embodiment 1, optionally, step 101 specifically includes:

Step 1011 , according to the first feature data and the preset random forest RF model, determine the target feature importance score of each feature class in the first feature data in the RF model.

Step 1012 , according to the target feature importance score of each feature class in the RF model, determine the target feature class related to the target service.

Optionally, step 1011 may specifically include:

Step 10111: Acquire respective first samples corresponding to at least two decision trees according to the first feature data.

Step 10112, for each decision tree, based on its corresponding first sample, determine the first feature importance score of m random feature classes among the M feature classes in the decision tree, where M is the first feature importance score in the first feature data. The total number of feature classes included, m is an integer less than M.

Step 10113, for each feature class, determine the second feature importance score of the feature class based on the first feature importance score of the feature class in each decision tree, and the second feature importance score is the feature class in each decision tree. The sum of the first feature importance scores in the decision tree.

Step 10114: Normalize the second feature importance scores of each feature class to obtain the target feature importance score of each feature class.

Optionally, for each decision tree, based on its corresponding first sample, determine the first feature importance score of m random feature classes in the decision tree among the M feature classes, including:

For each node in the decision tree, calculate the Gini index of each feature class of the m feature classes at the node;

For each feature class in the m feature classes, based on the Gini index of each node of the feature class in the decision tree, calculate and obtain the variation of the Gini index of the feature class at each node;

Based on the variation of the Gini index of each of the m feature classes at each node, the first feature importance score of the m feature classes in the decision tree is calculated and obtained.

Specifically, in the random forest algorithm, it is usually necessary to set up multiple decision trees, and each decision tree may include multiple nodes.

For each decision tree, its corresponding first sample is randomly selected from the first feature data. The first sample includes M feature classes, and the first sample is consistent with the feature classes included in the first feature data.

Exemplarily, for example, there are 100,000 pieces of first feature data, each with 800-dimensional features (that is, including M=800 feature classes), and 60,000*800-dimensional features can be randomly selected with replacement as the corresponding decision tree. first sample. For each decision tree, 60,000*800 dimensions are randomly selected. Obtain the first sample corresponding to each decision tree.

After the first sample corresponding to each decision tree is obtained, for each decision tree, based on its corresponding first sample, determine the number of random m feature classes among the M feature classes in the decision tree. A feature importance score, M is the total number of feature classes included in the first feature data, and m is an integer smaller than M.

Exemplarily, 500 feature classes are randomly selected from 800 feature classes, the decision of each node on the decision tree is determined based on these feature classes, and the optimal splitting method of nodes is determined according to these 500 feature classes.

Let VIM represent the feature importance score, GI represent the Gini index, and calculate the Gini index GIs of each feature class on the node s in the decision tree as:

Among them, K represents that under this feature class, there are K categories of user feature values, and p _sk represents the proportion of category k in the user feature value in node s.

Exemplarily, for feature class A, the current node s has 60 samples, of which some users browse for 1 day, some users browse for 2 days, and some users browse for 3 days. Under this feature class A, the samples are divided into For K=2 categories, count the total number of users who browsed for 1 day and 2 days, and the total number of users who browsed for 3 days. If there are 10 records in 1 day and 20 records in 2 days, then the users who browsed for 1 day and 2 days The total number is 30, and 30 have been browsed for 3 days, then the Gini index of feature class A on node s is:

After obtaining the Gini index of each node in the decision tree of each feature class in the m feature classes, for each feature class in the m feature classes, based on the Gini index of the feature class in each node in the decision tree, calculate Obtain the variation of the Gini index of the feature class at each node. The variation of the Gini index before and after the specific node s is divided into:

Among them, j represents the jth feature class, and GIl and GIr represent the Gini indices of the left branch node and the branch node of the jth feature class in section s, respectively.

According to this calculation, the variation of Gini index of each feature class at each node is obtained. Then, based on the variation of the Gini index of each feature class in the m feature classes at each node, the first feature importance score of the m feature classes in the decision tree is calculated and obtained. The first feature importance score of the jth feature class in this decision tree is:

Among them, i represents the ith decision tree, S represents the set of non-leaf nodes on the ith decision tree, and s represents the non-leaf nodes on the ith decision tree.

If there are n trees in the RF model, the obtained second feature importance score of the jth feature class is:

The second feature importance score is normalized to obtain the target feature importance score of each feature class. The target feature importance score for the jth feature class is:

Among them, M represents the total number of feature classes, and t represents the t-th feature class.

After obtaining the target feature importance score of each feature class, the target feature class related to the target business is determined according to the target feature importance score of each feature class in the RF model.

Optionally, according to the target feature importance score, a feature class with a score greater than a preset threshold is obtained as the target feature class. Or according to the target feature importance score, a preset proportion or a preset number of feature classes are obtained as the target feature classes. The specific selection manner may be set according to actual requirements, which is not limited in this embodiment.

As another implementable manner, on the basis of the above-mentioned first embodiment, optionally, the target business is a rental business, and the first feature data includes rental behavior feature data, and non-identity data mined based on the rental behavior feature data. Rental behavior characteristic data.

As another implementable manner, on the basis of the above-mentioned Embodiment 1, optionally, the user to be authenticated is authenticated based on the relevant data of the user to be authenticated, including:

Based on the relevant data of the user to be authenticated and the pre-trained classification model, the user to be authenticated is authenticated.

Optionally, the classification model is an Xgboost model; before the user to be authenticated is authenticated based on the relevant data of the user to be authenticated and the pre-trained classification model, the method further includes:

Step 2011, acquiring training sample data.

Step 2012, using the training sample data to train the pre-established Xgboost network to obtain an Xgboost model.

Specifically, the training sample data includes training feature data and labeling data. Among them, the training feature data includes user data of the target feature class, and the labeling data includes whether each piece of user data has been converted, that is, whether the user has rented a house. The training feature data that has actually been transformed is used as the positive sample set Y, and the negative sample set is extracted according to the amount of Y data according to a certain ratio, such as 1:1 or 1:2, that is, the training feature data that has not been transformed. Negative sample sets can use stratified random sampling. The positive sample set and the negative sample are set into a training sample set (ie, training sample data), and the pre-established Xgboost network is trained by using the training sample data to obtain an Xgboost model. The specific training process is consistent with the existing training process and will not be repeated here.

Optionally, the training sample data can be randomly divided according to a certain ratio, for example, randomly divided at a ratio of 3:7, part of which is used as a training sample set, and a part of which is used as a test set. Check the accuracy, AUC, and F1 value of the Xgboost model. Wait, and optimize to a reasonable range to obtain a trained Xgboost model to authenticate the user to be authenticated, and use the data classified as 1 as the target potential user, that is, the user to be authenticated who has passed the authentication.

AUC: the area under the curve, this curve is the ROC curve, the full name is the Receiver OperatingCharacteristic curve, it was first used in the analysis of electrical signals in the 1950s, and became popular in 1978 with the "Basic Principles of ROC Analysis" stand up. True Positive Rate (TPR, the ratio of the number of positive samples classified correctly by the classifier to the total number of positive samples) and False Positive Rate (FPR, the number of negative samples classified incorrectly by the classifier accounted for the total number of negative samples) of the ROC curve descriptor ratio of numbers).

F1 value: It is the harmonic mean of precision and recall, and its value is closer to the smaller of precision and recall.

Accuracy rate: refers to the proportion of the number of positive samples classified correctly by the classifier to the total number of positive samples classified by the classifier.

Recall rate: refers to the proportion of the number of correct positive samples classified by the classifier to the total number of positive samples

As shown in FIG. 3 , it is a schematic diagram of the overall authentication process provided in this embodiment. The feature extraction part is the process of obtaining useful features of potential users. The feature engineering part is the process of using the RF model to determine the target feature class, and the feature importance is to determine the importance of each feature class according to the target feature importance score, and determine a batch of more important target feature classes. The core feature set is the training sample data used to train the classification model, and the optimized model is the trained classification model, the prediction set, and the potential users identified by the optimized model.

It should be noted that, each implementable manner in this embodiment may be implemented independently, or may be implemented in combination in any combination under the condition of no conflict, which is not limited in this application.

In the user authentication method provided by this embodiment, the target feature class related to the target service is determined from the numerous feature classes of the user group by adopting a preset model, the feature of the target service is expanded, and further based on the expanded target feature class, Obtain the relevant data of the user to be authenticated according to the identification of the user to be authenticated, authenticate the user to be authenticated based on the relevant data of the user to be authenticated, and push the content related to the target service to the user to be authenticated that has passed the authentication, improving the It can effectively improve the accuracy of business information push, thereby improving the marketing effect.

Embodiment 3

This embodiment provides an authentication device for a user, which is used to execute the method of the first embodiment.

As shown in FIG. 4 , it is a schematic structural diagram of a user authentication apparatus provided in this embodiment. The user authentication device 30 includes a determination module 31 , a reception module 32 , an acquisition module 33 , an authentication module 34 and a push module 35 .

The determining module is configured to determine the target feature class related to the target service based on the first preset sample data, where the first preset sample data includes the first feature data of the user's historical behavior; the receiving module is configured to receive an authentication request , the authentication request includes the identification of the user to be authenticated; the acquisition module is used to obtain the relevant data of the user to be authenticated according to the identification of the user to be authenticated, and the relevant data includes the second feature of each target feature class corresponding to the user to be authenticated data; the authentication module is used to authenticate the user to be authenticated based on the relevant data of the user to be authenticated; the push module is used to push the content related to the target service to the user to be authenticated if the authentication is passed.

Regarding the apparatus in this embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be described in detail here.

According to the user authentication device provided in this embodiment, the target feature class related to the target service is determined from the numerous feature classes of the user group by adopting a preset model, the feature of the target service is expanded, and further based on the expanded target feature class , obtain the relevant data of the user to be authenticated according to the identity of the user to be authenticated, authenticate the user to be authenticated based on the relevant data of the user to be authenticated, and push the content related to the target service to the user to be authenticated who has passed the authentication, The mining volume and accuracy of potential users are improved, and the accuracy of business information push can be effectively improved, thereby improving the marketing effect.

Embodiment 4

This embodiment provides further supplementary descriptions of the devices provided in the foregoing embodiments.

As an implementable manner, on the basis of the foregoing embodiment, optionally, the determination module is specifically used for:

According to the first feature data and the preset random forest RF model, determine the target feature importance score of each feature class in the first feature data in the RF model;

According to the target feature importance score of each feature class in the RF model, the target feature class related to the target business is determined.

Optionally, determine a module, specifically for:

According to the first feature data, obtain the first samples corresponding to at least two decision trees respectively;

For each decision tree, based on its corresponding first sample, determine the first feature importance score of m random feature classes among the M feature classes in the decision tree, where M is the feature included in the first feature data The total number of classes, m is an integer less than M;

For each feature class, based on the first feature importance score of the feature class in each decision tree, the second feature importance score of the feature class is determined, and the second feature importance score is the feature class in each decision tree. The sum of the first feature importance scores of ;

The second feature importance score of each feature class is normalized to obtain the target feature importance score of each feature class.

Optionally, determine a module, specifically for:

For each node in the decision tree, calculate the Gini index of each feature class of the m feature classes at the node;

For each feature class in the m feature classes, based on the Gini index of each node of the feature class in the decision tree, calculate and obtain the variation of the Gini index of the feature class at each node;

Based on the variation of the Gini index of each of the m feature classes at each node, the first feature importance score of the m feature classes in the decision tree is calculated and obtained.

In some embodiments, optionally, the determining module is specifically used for:

According to the importance score of the target feature, the feature class with the score greater than the preset threshold is obtained as the target feature class.

As another implementable manner, on the basis of the above-mentioned embodiment, optionally, the target business is a rental business, and the first feature data includes rental behavior feature data and non-rental properties mined based on the rental behavior feature data. Class behavior characteristic data.

As another implementable manner, on the basis of the foregoing embodiment, optionally, the authentication module is specifically used for:

Based on the relevant data of the user to be authenticated and the pre-trained classification model, the user to be authenticated is authenticated.

Optionally, the classification model is an Xgboost model; the acquisition module is also used for:

Get training sample data;

The pre-established Xgboost network is trained with the training sample data to obtain the Xgboost model.

Regarding the apparatus in this embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be described in detail here.

It should be noted that, each implementable manner in this embodiment may be implemented independently, or may be implemented in combination in any combination under the condition of no conflict, which is not limited in this application.

According to the user authentication device of the present embodiment, the target feature class related to the target service is determined from the numerous feature classes of the user group by adopting a preset model, the feature of the target service is expanded, and further based on the expanded target feature class, Obtain the relevant data of the user to be authenticated according to the identification of the user to be authenticated, authenticate the user to be authenticated based on the relevant data of the user to be authenticated, and push the content related to the target service to the user to be authenticated that has passed the authentication, improving the It can effectively improve the accuracy of business information push, thereby improving the marketing effect.

Embodiment 5

This embodiment provides a computer device for executing the method provided by the foregoing embodiment.

As shown in FIG. 5 , it is a schematic structural diagram of a computer device provided in this embodiment. The computer device 50 includes: at least one processor 51 and a memory 52;

The memory stores a computer program; at least one processor executes the computer program stored in the memory to implement the methods provided by the above embodiments.

According to the computer equipment of the present embodiment, the target feature class related to the target service is determined from the numerous feature classes of the user group by using the preset model, the features of the target service are expanded, and further based on the expanded target feature class, according to the target feature class to be identified The identification of the authorized user obtains the relevant data of the user to be authenticated, authenticates the user to be authenticated based on the relevant data of the user to be authenticated, and pushes the content related to the target service to the user to be authenticated who has passed the authentication. The mining volume and accuracy can effectively improve the accuracy of business information push, thereby improving the marketing effect.

Embodiment 6

This embodiment provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed, the method provided by any of the foregoing embodiments is implemented.

According to the computer-readable storage medium of the present embodiment, the target feature class related to the target service is determined from the numerous feature classes of the user group by using a preset model, the feature of the target service is expanded, and further based on the expanded target feature class, Obtain the relevant data of the user to be authenticated according to the identification of the user to be authenticated, authenticate the user to be authenticated based on the relevant data of the user to be authenticated, and push the content related to the target service to the user to be authenticated that has passed the authentication, improving the It can effectively improve the accuracy of business information push, thereby improving the marketing effect.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution in this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware, or can be implemented in the form of hardware plus software functional units.

The above-mentioned integrated units implemented in the form of software functional units can be stored in a computer-readable storage medium. The above-mentioned software function unit is stored in a storage medium, and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) to execute the methods described in the various embodiments of the present application. some steps. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program codes .

Those skilled in the art can clearly understand that, for the convenience and brevity of the description, only the division of the above functional modules is used for illustration. The internal structure is divided into different functional modules to complete all or part of the functions described above. For the specific working process of the apparatus described above, reference may be made to the corresponding process in the foregoing method embodiments, and details are not described herein again.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions described in the foregoing embodiments can still be modified, or some or all of the technical features thereof can be equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present application. scope.

Claims (10)

1. A method for authenticating a user, comprising:

determining a target feature class related to a target service based on first preset sample data, wherein the first preset sample data comprises first feature data of user historical behaviors;

receiving an authentication request, wherein the authentication request comprises an identifier of a user to be authenticated;

acquiring related data of the user to be authenticated according to the identifier of the user to be authenticated, wherein the related data comprises second feature data of each target feature class corresponding to the user to be authenticated;

authenticating the user to be authenticated based on the related data of the user to be authenticated;

and if the authentication is passed, pushing the content related to the target service to the user to be authenticated.

2. The method according to claim 1, wherein the determining a target feature class related to a target service based on the first preset sample data comprises:

determining target feature importance scores of all feature classes in the first feature data in a preset random forest RF model according to the first feature data and the RF model;

and determining a target feature class related to the target service according to the target feature importance scores of the feature classes in the RF model.

3. The method of claim 2, wherein determining a target feature importance score for each feature class in the first feature data in the RF model based on the first feature data comprises:

obtaining at least two first samples corresponding to the decision trees respectively according to the first characteristic data;

for each decision tree, determining a first feature importance score of M random feature classes in the decision tree based on a corresponding first sample, wherein M is the total number of the feature classes included in the first feature data, and M is an integer smaller than M;

for each feature class, determining a second feature importance score of the feature class based on a first feature importance score of the feature class in each decision tree, wherein the second feature importance score is the sum of the first feature importance scores of the feature class in each decision tree;

and carrying out normalization processing on the second feature importance scores of the feature classes to obtain the target feature importance scores of the feature classes.

4. The method of claim 3, wherein determining, for each decision tree, a first feature importance score for a random M of the M feature classes in the decision tree based on its corresponding first sample comprises:

for each node in the decision tree, calculating the Gini index of each feature class in the m feature classes at the node;

for each feature class in the m feature classes, calculating and obtaining Gini index variable quantity of the feature class at each node based on the Gini index of each node in the decision tree of the feature class;

and calculating to obtain a first feature importance score of the m feature classes in the decision tree based on Gini index variation of each feature class in the m feature classes in each node.

5. The method of claim 2, wherein determining a target feature class associated with a target service based on the target feature importance scores of the feature classes in the RF model comprises:

and acquiring a feature class with the score larger than a preset threshold value as the target feature class according to the target feature importance score.

6. The method of claim 1, wherein the target service is a house renting service, wherein the first feature data comprises house renting behavior feature data, and non-house renting behavior feature data mined based on the house renting behavior feature data.

7. The method of claim 1, wherein the authenticating the user to be authenticated based on the data related to the user to be authenticated comprises:

and authenticating the user to be authenticated based on the related data of the user to be authenticated and a pre-trained classification model.

8. An authentication apparatus for a user, comprising:

the determining module is used for determining a target feature class related to a target service based on first preset sample data, wherein the first preset sample data comprises first feature data of user historical behaviors;

the receiving module is used for receiving an authentication request, wherein the authentication request comprises an identifier of a user to be authenticated;

the acquisition module is used for acquiring the related data of the user to be authenticated according to the identification of the user to be authenticated, wherein the related data comprises second feature data of each target feature class corresponding to the user to be authenticated;

the authentication module is used for authenticating the user to be authenticated based on the related data of the user to be authenticated;

and the pushing module is used for pushing the content related to the target service to the user to be authenticated if the authentication is passed.

9. A computer device, comprising: at least one processor and memory;

the memory stores a computer program; the at least one processor executes the memory-stored computer program to implement the method of any of claims 1-7.

10. A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, which computer program, when executed, implements the method of any one of claims 1-7.

Images