CN120471047A

CN120471047A - Insurance policy entry method, device, equipment and storage medium

Info

Publication number: CN120471047A
Application number: CN202510541377.6A
Authority: CN
Inventors: 黄昊
Original assignee: Ping An Property and Casualty Insurance Company of China Ltd
Current assignee: Ping An Property and Casualty Insurance Company of China Ltd
Priority date: 2025-04-27
Filing date: 2025-04-27
Publication date: 2025-08-12

Abstract

The present application belongs to the field of artificial intelligence technology, and discloses a method, apparatus, device, and storage medium for insurance policy entry, including: accessing a preset insurance database, and constructing an insurance knowledge graph based on the preset insurance database; obtaining initial order information, and converting the initial order information into text graph structure information; processing the text graph structure information using a graph neural network to generate intermediate order information; correcting the intermediate order information according to the insurance knowledge graph to obtain error information, and using a sequence-to-sequence model to convert the intermediate order information according to the error information to generate target order information. The present application can be applied to insurance policy entry scenarios such as finance and medical care, and improves the accuracy and efficiency of insurance policy entry.

Description

Policy entry method, device, equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence, and in particular, to a policy entry method, apparatus, device, and storage medium.

Background

Along with the rapid development of information, the online single-flow is gradually realized, the revolution greatly improves the single-flow efficiency of the service personnel, shortens the waiting time of the clients, does not need to go to the site for handling, and brings remarkable operation optimization for the insurance and finance industries;

However, in the conventional online bill discharging process, for example, in the step of entering a policy, a large amount of key data such as customer information, policy details and the like need to be manually input by a service operator, which results in time and labor consumption, and easy input errors caused by human factors in the conventional online bill discharging process, so that the complexity of subsequent business processing is increased, and even disputes in claims may be caused.

For example, in the financial field, many insurance companies start to implement an electronic policy system, when a customer purchases an insurance product, only information filling, checking, payment and other steps are needed to be completed through an official website of the insurance company or a cooperation channel (such as a bank) and the like, then a salesman manually inputs key data such as identity information, policy type, insurance period, amount and the like of the customer according to the filled information of the customer, and in the manual recording process of the salesman, the possibility of recording errors easily occurs, so that the subsequent insurance company cannot accurately recognize the customer information, or other claim disputes and the like are generated.

In summary, the existing online single-process has the disadvantages of lower policy entry efficiency and lower policy entry accuracy.

Disclosure of Invention

The embodiment of the application aims to provide a policy entry method, a policy entry device, a policy entry equipment and a storage medium, which mainly aim to improve the policy entry efficiency and the policy entry accuracy in a policy issuing process.

In order to solve the above technical problems, the embodiment of the present application provides a policy entry method, which adopts the following technical scheme:

Accessing a preset insurance database, and constructing an insurance knowledge graph according to the preset insurance database;

acquiring initial recording information, and converting the initial recording information into text diagram structure information;

processing the text graph structure information by using a graph neural network to generate intermediate record information;

and correcting the middle record information according to the insurance knowledge graph to obtain error information, and converting the middle record information by adopting a sequence-to-sequence model according to the error information to generate target record information.

In order to solve the above technical problems, the embodiment of the present application further provides a policy entry device, which adopts the following technical scheme:

The map construction module is used for accessing a preset insurance database and constructing an insurance knowledge map according to the preset insurance database;

The recording information acquisition module is used for acquiring initial recording information and converting the initial recording information into text diagram structure information;

The model processing module is used for processing the text graph structure information by using the graph neural network to generate intermediate record list information;

and the error information error correction module is used for correcting the intermediate record information according to the insurance knowledge graph to obtain error information, and converting the intermediate record information by adopting a sequence-to-sequence model according to the error information to generate target record information.

In order to solve the technical problem, the embodiment of the application further provides computer equipment, which comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions which can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the policy entry method.

In a fourth aspect, in order to solve the above technical problem, an embodiment of the present application further provides a computer readable storage medium storing a computer program, which when executed by a processor, implements the policy entry method as described above.

Compared with the prior art, the embodiment of the application has the following main beneficial effects:

The basis can be provided for the error correction of the follow-up recording information through the insurance knowledge graph constructed according to the preset insurance database, and the accuracy of recording the insurance policy is indirectly improved;

The method and the system have the advantages that the initial record information is converted into the text diagram structural information, the identification of a follow-up graphic neural network is facilitated, the accuracy of insurance record is improved, the computer is facilitated to process rapidly through the structured text diagram structural information, and the efficiency of insurance record information flow is improved.

The method has the advantages that the intermediate record list information is generated by utilizing the graphic neural network, the text graphic structure information can be automatically processed, the manual intervention and the error rate are reduced, the key information in the graphic structure is automatically extracted, the automatic entry of the insurance policy is realized, the insurance policy entry efficiency is improved, the customer satisfaction is improved, the graphic structure data can be directly processed through the graphic neural network, and the insurance policy entry accuracy is improved through information transmission and parameter updating among nodes.

The insurance knowledge graph can integrate data with different sources and different structures, error correction is carried out through the insurance knowledge graph, errors of intermediate recording information can be accurately identified, the error correction accuracy of the insurance knowledge graph is improved, and further the accuracy of insurance recording is improved.

Drawings

In order to more clearly illustrate the solution of the present application, a brief description will be given below of the drawings required for the description of the embodiments of the present application, it being apparent that the drawings in the following description are some embodiments of the present application, and that other drawings may be obtained from these drawings without the exercise of inventive effort for a person of ordinary skill in the art.

FIG. 1 is an exemplary system architecture diagram in which the present application may be applied;

FIG. 2 is a flow chart of one embodiment of a policy entry method according to the application;

FIG. 3 is a schematic diagram of the structure of one embodiment of a policy entry device according to the application;

Fig. 4 is a schematic structural view of an embodiment of the device according to the application.

Detailed Description

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs, the terms used in the description herein are used for the purpose of describing particular embodiments only and are not intended to limit the application, and the terms "comprising" and "having" and any variations thereof in the description of the application and the claims and the above description of the drawings are intended to cover non-exclusive inclusions. The terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

In order to make the person skilled in the art better understand the solution of the present application, the technical solution of the embodiment of the present application will be clearly and completely described below with reference to the accompanying drawings.

As shown in fig. 1, the system architecture 100 may include a terminal device 101, a network 102, and a server 103, where the terminal device 101 may be a notebook 1011, a tablet 1012, or a cell phone 1013. Network 102 is the medium used to provide communication links between terminal device 101 and server 103. Network 102 may include various connection types such as wired, wireless communication links, or fiber optic cables.

A user may interact with the server 103 via the network 102 using the terminal device 101 to receive or send messages or the like. Various communication client applications, such as a web browser application, a shopping class application, a search class application, an instant messaging tool, a mailbox client, social platform software, and the like, may be installed on the terminal device 101.

The terminal device 101 may be various electronic devices having a display screen and supporting web browsing, and the terminal device 101 may be an electronic book reader, an MP3 player (Moving Picture Experts Group Audio Layer III, moving picture experts compression standard audio layer III), an MP4 (Moving Picture Experts Group Audio Layer IV, moving picture experts compression standard audio layer IV) player, a laptop portable computer, a desktop computer, or the like, in addition to the notebook 1011, the tablet 1012, or the mobile phone 1013.

The server 103 may be a server providing various services, such as a background server providing support for pages displayed on the terminal device 101.

It should be noted that, the policy entry method provided by the embodiment of the present application is generally executed by a server/terminal device, and accordingly, the policy entry device is generally disposed in the server/terminal device.

It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

With continued reference to fig. 2, a flow chart of one embodiment of a policy entry method according to the application is shown. The order of the steps in the flowchart may be changed and some steps may be omitted according to various needs. The policy entry method provided by the embodiment of the application can be applied to any scene needing policy entry, and can be applied to products of the scenes. The policy entry method comprises the following steps:

Step S201, accessing a preset insurance database, and constructing an insurance knowledge graph according to the preset insurance database.

In this embodiment, the electronic device (for example, the server/terminal device shown in fig. 1) on which the policy entry method operates may be an insurance database preset by a wired connection manner or a wireless connection manner, and the above-mentioned initial form recording information. It should be noted that the wireless connection may include, but is not limited to, 3G/4G/5G connection, wiFi connection, bluetooth connection, wiMAX connection, zigbee connection, UWB (ultra wideband) connection, and other now known or later developed wireless connection.

In this embodiment, the preset insurance database refers to a database storing a large amount of insurance-related data and information, including insurance product information (such as name, type, guarantee scope, premium, insurance amount, etc. of the insurance product), customer information (such as customer name, age, sex, contact, occupation, and insurance purchase history of the customer, claim records, etc.), claim data (including detailed information of all claim cases, such as claim types, claim amounts, etc.), insurance terms and legal regulations (such as term content of various insurance products, and related insurance law regulations and regulatory policies), industry reports and statistics (including market analysis reports, statistics, etc. of the insurance industry), and after accessing the preset insurance database, acquiring the insurance data information of the insurance database, and constructing an insurance knowledge graph according to the insurance data information.

In one embodiment, the building an insurance knowledge graph according to the preset insurance database includes:

Acquiring insurance data information from the preset insurance database, and performing entity identification on the insurance data information to obtain at least one map entity;

analyzing the semantic relation of each map entity, and extracting the relation of at least one map entity according to the semantic relation to obtain a map relation;

And constructing the insurance knowledge graph by taking the graph entity as a node and the graph relation as an edge.

In this embodiment, required insurance data information (such as insurance product information and insurance claim information) related to insurance is extracted from a preset insurance database, the extracted insurance data information is preprocessed, the preprocessing includes but is not limited to data cleaning of the insurance data information, repeated, wrong and invalid data are removed, missing values and converted data formats are filled, the entity types to be identified are definitely determined according to preset entity tables, such as insurance products and claim settlement processes, then entity identification is performed on the insurance data information by adopting a natural language processing technology (such as named entity identification (NER)), the entity to be identified is extracted from the insurance data information, the extracted entity is cleaned, repeated or wrong entity is removed, at least one map entity is obtained, semantic relations existing among the map entities are defined according to a preset relation table, relation extraction is performed according to the preset relation table, the map entity is finally used as a node of the map, the map knowledge relationship is used as a side of the map, and the knowledge of the insurance is constructed.

In the embodiment, the basis can be provided for the error correction of the follow-up recording information through the insurance knowledge graph constructed according to the preset insurance database, and the accuracy of recording the insurance policy is indirectly improved.

Step S202, obtaining initial recording information, and converting the initial recording information into text diagram structure information.

In this embodiment, the initial form information refers to a voice description or text description of the insurance form information by the service personnel, including basic information (such as name, contact way, address, etc.) of the customer, detailed information (such as insurance product type, insurance product name, insurance fee, etc.) of the insurance product, related additional information (such as application time, effective date, etc.), and after receiving the initial form information uploaded by the service personnel, preprocessing the initial form information, where the preprocessing includes word segmentation, part-of-speech labeling, and named entity identification, where the named entity identification includes but is not limited to identification of person name, place name, product name, etc. in the text, then performing grammar relation analysis on the initial form information after the preprocessing, and converting the initial form information into text-graph structural information according to the parsed grammar relation.

In one embodiment, after the acquiring the initial transcript information, the method further comprises:

Identifying the type of the initial record information;

and when the type of the initial record information is recognized as a voice type, converting the initial record information into a text form by utilizing a voice conversion technology.

In this embodiment, after receiving initial recording information uploaded by a salesman, the type of the initial recording information is identified by identifying the file extension of the initial recording information, when the extension of the initial recording information is identified as wav, the type of the initial recording information is determined to be an audio type, and the initial recording information is converted into a Text form by using a Speech conversion technology (such as a Speech-to-Text API).

In the embodiment, by paying out the initial recording information input in various forms, convenience and flexibility of inputting by the service personnel are improved, efficiency of recording the information in subsequent insurance is improved, and experience of using by the service personnel is also improved.

In one embodiment, the converting the initial transcript information into text-map structure information includes:

performing text preprocessing operation on the initial list information, wherein the text preprocessing operation comprises word segmentation, stop word removal and part-of-speech tagging;

Based on the initial record information after the text preprocessing operation, adopting a named entity recognition technology to recognize the initial record information to obtain a plurality of text entities;

performing relationship identification on a plurality of text entities through a preset natural language processing tool to obtain entity relationships of the plurality of text entities;

And constructing the text graph structure information by taking a plurality of text entities as nodes of the text graph structure and taking a plurality of entity relations as edges.

In this embodiment, the text preprocessing operation includes word segmentation, stop word removal and part-of-speech tagging, and the text preprocessing operation includes word segmentation processing on the initial list information by using a word segmentation tool (such as jieba word segmentation, etc.), segmenting continuous initial list information into independent words, filtering stop words in the initial list information according to a preset stop word list, such as "having" or the like, finally part-of-speech tagging on the segmented initial list information by using a word-tagging tool, completing the text preprocessing operation on the initial list information, recognizing a plurality of text entities (such as name and insurance name, etc.) in the initial list information by using a named entity recognition tool (such as a NER module of jieba), after recognizing a plurality of text entities in the initial list information, configuring a plurality of text entity models according to a plurality of syntactic entity relationships (such as a text structure map, a plurality of text entity relationships, and the like), and using a plurality of text entity relationships as a text structure map, and a text structure relationship as a text structure relationship.

In the embodiment, the definition of the initial insurance policy record information can be improved by carrying out entity identification and entity relationship identification on the initial policy record information and constructing text diagram structure information based on the identified entity and entity relationship, a basis is provided for the follow-up insurance policy record information input, and the accuracy of the policy record flow is improved.

In the financial field, as an implementation example A, a salesman receives information of client's name three, 35 years of life from an online platform of a insurance company, the contact telephone is 138xxxx8888, I want to purchase a major disease insurance, the insurance is 200 ten thousand yuan, the salesman inputs the information of three input into the system as initial form recording information in the form of dictation or text input, and performs text preprocessing operation on the initial form recording information, i (r)/name (v)/three (PER)/and (w)/this year (TIME)/35 years of age (m)/and (w)/contact telephone (n)/is (v)/138 xxxx8888 (m)/v. (w)/me (r)/want (v)/purchase (v)/share (m)/major disease insurance (nz)/(w)/insurance (n)/200 ten thousand yuan (m). And (w), identifying by adopting a named entity identification technology based on the preprocessed initial list information to obtain a text entity corresponding to the initial list information, wherein Zhang San is a name of a person, the age of the person is 35 years old, 138xxxx8888 is a telephone number, the number of the person is one, serious disease insurance is an insurance product name, the insurance amount of the person is 200 ten thousand yuan, analyzing relations among the entities through a natural language processing tool, such as the relation between Zhang San and 35 years old, taking the text entity as a node, the relation among the text entities as an edge, and converting the initial list information into text graph structure information.

In the embodiment, the initial record information is converted into the text diagram structural information, so that the identification of a subsequent diagram neural network is facilitated, the accuracy of the insurance record is improved, the computer is facilitated to rapidly process the structured text diagram structural information, and the efficiency of the insurance record information flow is improved.

And step S203, processing the text graph structure information by using a graph neural network to generate intermediate record information.

In this embodiment, the aforementioned graph neural network refers to a deep learning model GNN (Graph Neural Network), and focuses on processing graph structure data, where the graph neural network includes an input layer, a multi-layer graph convolution layer, and an output layer, the intermediate recording information is standardized recording information generated according to initial recording information, the text graph structure information is processed through the graph neural network to generate standardized intermediate recording information, and the intermediate recording information is stored in a preset intermediate database for subsequent extraction and processing.

In one embodiment, the processing the text graph structure information by using the graph neural network to generate intermediate transcript information includes:

identifying nodes of the text graph structure information, extracting feature vectors of each node, and obtaining initial feature vectors corresponding to each node;

inputting all the initial feature vectors into the graph neural network through an input layer of the graph neural network;

Based on each initial feature vector, collecting neighbor node information of the initial feature vector through a graph convolution layer of the graph neural network, and fusing the initial feature vector and the neighbor node information to obtain a target feature vector;

And converting the target feature vector into structured data through an output layer of the graph neural network to obtain the intermediate record information.

In this embodiment, all nodes are identified from text graph structure information, each node is converted into a feature vector through a vector conversion technology (such as word2Vec, etc.), an initial feature vector corresponding to each node is obtained, all initial feature vectors are formed into a matrix based on all initial feature vectors to obtain an input matrix, the input matrix is input into a graph neural network through an input layer of the graph neural network, wherein the input layer is a simple linear transformation layer used for mapping the input matrix into an internal representation space of the graph neural network, for the initial feature vector corresponding to each node, information of neighbor nodes is collected through the graph convolution layer, the neighbor nodes refer to nodes with connection relation with the node in the input matrix, the collected information of the neighbor nodes is fused with the initial feature vectors, the fusion mode includes, but is not limited to, weighted average, summation, etc., the initial feature vectors corresponding to each node are processed through stacking a plurality of graph convolution layers, finally the target feature vector is obtained, and intermediate list information is output through an output layer of the graph neural network.

In the embodiment, the complex relation of the text graph structural information can be effectively captured through the graph neural network, the initial feature vector of each node is fused with the neighbor node information entry, so that the extracted features are more comprehensive and accurate, the information extraction accuracy is improved, the output layer of the graph neural network converts the target feature vector into the structured data, the intermediate form information is obtained, and the subsequent processing is more convenient and efficient.

In one embodiment, after the text graph structure information is processed by using the graph neural network to generate intermediate transcript information, the method further comprises:

analyzing the intermediate list information to obtain main risk types and insured features;

Screening a first additional risk set related to the main risk type according to the main risk type based on the insurance knowledge graph;

screening from the first additional risk set according to the features of the insured person to obtain a second additional risk set;

Sorting the second additional risk sets according to the monetary factor of each additional risk in the second additional risk sets, selecting the additional risk sets with preset ranks from the sorted second additional risk sets as target additional risk sets, and pushing the target additional risk sets.

In this embodiment, the middle record information is parsed, key information such as a main risk type and insured person features in the middle record information is extracted, the main risk type refers to insurance types in the middle record information, such as car risk, health risk, life risk and the like, the insured person features refer to age, gender, health condition and occupation, the first additional risk set is screened out by taking the main risk type as a starting node according to a relation established in the insurance knowledge map based on an insurance knowledge map, the first additional risk set is further screened out according to the insured person features on the basis of the first additional risk set, for example, the main risk type is medical risk, then for insured persons with older ages, such as female specific diseases risk can be screened out, and a first additional risk set is obtained.

In the embodiment, the main risk type and the insured person characteristic are obtained by analyzing the intermediate form recording message, personalized additional risk pushing can be carried out according to the specific situation of the insured person, the accuracy of the additional risk pushing is improved, the additional risk related to the main risk type can be rapidly screened out by pushing through the insurance knowledge graph, the manual screening time is reduced, and the efficiency of the additional risk recommendation is improved.

Following example a, the text structured information is processed through a neural network to generate intermediate transcript information, which may exist in a structured data format:

And screening out the target additional risk based on the data, and adding the target additional risk to the intermediate list information after the target additional risk is selected by the user.

In the embodiment, the intermediate record information is generated by utilizing the graphic neural network, so that the text graphic structure information can be automatically processed, the manual intervention and the error rate are reduced, the key information in the graphic structure is automatically extracted, the automatic entry of the policy is realized, the policy entry efficiency is improved, the customer satisfaction is improved, the graphic structure data can be directly processed by utilizing the graphic neural network, and the accuracy of the policy entry is improved by information transmission and parameter updating among nodes.

And S204, correcting the intermediate record information according to the insurance knowledge graph to obtain error information, and converting the intermediate record information by adopting a sequence-to-sequence model according to the error information to generate target record information.

In this embodiment, the middle record information is corrected according to the insurance knowledge graph, the middle record information is input into the insurance knowledge graph, the accuracy of the middle record information is verified by utilizing nodes and edges in the graph, when errors or omission exists in the middle record information, if an insurance product selected by an insured person is not in compliance with the regulations, the filled insurance cost is not in compliance with the insurance product regulations, the middle record information is marked in type to obtain error information, the middle record information is replaced by adopting a sequence-to-sequence model according to the error information, when the replacement is completed, target record information is generated, the recorded record information is fed back to a service staff to be finally determined, after confirming that the error exists, a formal insurance file is generated according to the recorded data, and is sent to a client for signing and archiving, wherein the sequence-to-sequence (Seq 2 Seq) model is a depth model for processing sequence data, and can convert an input sequence with an indefinite length into an output sequence with an indefinite length.

In one embodiment, before said error correction of said intermediate transcript information according to said insurance knowledge-graph, the method further comprises:

Correcting the middle record information through a dictionary to obtain the middle record information corrected by the dictionary;

calculating the intermediate form information subjected to dictionary error correction through a random forest model, and calculating to obtain the probability of an error statement;

and when the probability of the error statement exceeds a preset error threshold value, marking the error statement to obtain the error information.

In this embodiment, the intermediate list information is primarily corrected by a dictionary, the intermediate list information is corrected by spelling and grammar, and is replaced by correct spelling information in the dictionary, so as to obtain the intermediate list information after dictionary correction, a random forest model is used to calculate the intermediate list information after dictionary correction, the probability of an error sentence in the intermediate list information after dictionary correction is calculated, specifically, the intermediate list information is subjected to text vectorization, the intermediate list information is converted into a numerical feature by Word Bag model (Bag of Words), TF-IDF (Word frequency-inverse document frequency) or Word embedding (Word 2Vec, BERT, etc.), key features are extracted from the numerical feature, the key features include but are not limited to vocabulary features (such as frequency, word part, etc.), grammar features (such as sentence structure, dependency relation, etc.), parameters of the random forest model are set, such as the number of decision trees, the maximum depth, the minimum sample number, etc., the key features are input into the random forest model which is trained, the probability of the error sentence corresponding to the keyword sentence is calculated according to the input key features, and the error sentence is marked when the probability of the error sentence is large, the error sentence is obtained.

In this embodiment, the error information obtained by the random forest model and the error information obtained by the insurance knowledge graph are converted from the sequence to the sequence model.

In the embodiment, the comprehensiveness of error correction can be improved through double error correction of the dictionary and the random forest model, and the accuracy of policy entry is further improved.

In one embodiment, the converting the intermediate transcript information using a sequence-to-sequence model according to the error information to generate target transcript information includes:

Identifying the error type according to the type mark of the error information and the position of the error information in the intermediate recording information;

coding the error information through an encoder from the sequence to a sequence model to obtain error information codes;

And acquiring context information according to the position of the error information in the intermediate recording information, and decoding the error information through the sequence-to-sequence decoder according to the context information and the error type to generate the target recording information.

In this embodiment, the error type of the error information and the position of the error information in the intermediate recording information are determined according to the type mark when the error information is identified, the error information is input into a sequence-to-sequence model, the error information is encoded into a vector representation by an encoder of the sequence-to-sequence model, then the text before and after the error information is extracted from the intermediate recording information according to the position of the error information and a predefined length as context information, the error encoded information and the context information are transmitted to a decoder, the decoder predicts the character or word sequence of the generated target recording information according to the error encoded information and the context information, and the decoder outputs the complete target recording information after all steps are completed.

In the embodiment, the errors in the intermediate form information can be automatically processed through the sequence-to-sequence model without manual checking and correction, so that the efficiency of entering the form is improved, the sequence-to-sequence model can accurately identify and correct the errors in the intermediate form information by giving context information and error types, the erroneous judgment of manual correction is reduced, and the accuracy of entering the form is improved.

And (3) receiving the practical example A, correcting the middle record information in the form of structured data through a dictionary, a random forest model and an insurance knowledge graph, when error information exists, replacing the error statement through a sequence-to-sequence model to generate complete target record information, feeding the target record information back to a salesman for confirmation, and finishing record entry after the salesman confirms no error.

In the embodiment, the insurance knowledge graph can integrate data with different sources and different structures, error correction is carried out through the insurance knowledge graph, errors of intermediate recording information can be accurately identified, the error correction accuracy of the insurance knowledge graph is improved, and further the accuracy of the insurance recording is improved.

It should be emphasized that, to further ensure the privacy and security of the aforementioned recording information, etc., the aforementioned recording information, etc. may also be stored in a node of a blockchain.

The blockchain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm and the like. The blockchain (Blockchain), essentially a de-centralized database, is a string of data blocks that are generated in association using cryptographic methods, each of which contains information from a batch of network transactions for verifying the validity (anti-counterfeit) of its information and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like.

The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Wherein artificial intelligence (ARTIFICIALINTELLIGENCE, AI) is the theory, method, technique, and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend, and expand human intelligence, sense the environment, acquire knowledge, and use knowledge to obtain optimal results.

Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by computer readable instructions stored in a computer readable storage medium that, when executed, may comprise the steps of the embodiments of the methods described above. The storage medium may be a nonvolatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a random access Memory (Random Access Memory, RAM).

It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.

With further reference to fig. 3, as an implementation of the method shown in fig. 2, the present application provides an embodiment of a policy entry apparatus, which corresponds to the method embodiment shown in fig. 2, and which is particularly applicable to various computer devices.

As shown in fig. 3, the policy entry device 300 in this embodiment includes a map construction module 301, a recording information acquisition module 302, a model processing module 303, and an error information correction module 304. Wherein:

the map construction module 301 is configured to access a preset insurance database, and construct an insurance knowledge map according to the preset insurance database;

in one embodiment, the map construction module comprises:

The entity identification sub-module is used for acquiring insurance data information from the preset insurance database, and carrying out entity identification on the insurance data information to obtain at least one map entity;

the relation extraction sub-module is used for analyzing the semantic relation of each map entity, and extracting the relation of at least one map entity according to the semantic relation to obtain a map relation;

and the map construction submodule is used for constructing the insurance knowledge map by taking the map entity as a node and taking the map relation as an edge.

The recording information acquisition module 302 is configured to acquire initial recording information, and convert the initial recording information into text diagram structure information;

In one embodiment, the recording information acquisition module includes:

The preprocessing sub-module is used for performing text preprocessing operation on the initial recording information, wherein the text preprocessing operation comprises word segmentation, stop word removal and part-of-speech tagging;

The entity identification sub-module is used for identifying the initial record information by adopting a named entity identification technology based on the initial record information after the text preprocessing operation to obtain a plurality of text entities;

the entity relation recognition sub-module is used for carrying out relation recognition on a plurality of text entities through a preset natural language processing tool to obtain entity relations of the plurality of text entities;

And the diagram structure information construction sub-module is used for constructing the text diagram structure information by taking a plurality of text entities as nodes of the text diagram structure and taking a plurality of entity relations as edges.

The model processing module 303 processes the text graph structure information by using a graph neural network to generate intermediate record list information;

In one embodiment, the model processing module includes:

the node characteristic extraction sub-module is used for identifying nodes of the text graph structural information, extracting characteristic vectors of each node and obtaining initial characteristic vectors corresponding to each node;

The model input sub-module is used for inputting all the initial feature vectors into the graph neural network through an input layer of the graph neural network;

the graph convolution sub-module is used for collecting neighbor node information of the initial feature vector through a graph convolution layer of the graph neural network based on each initial feature vector, and fusing the initial feature vector and the neighbor node information to obtain a target feature vector;

And the feature conversion sub-module is used for converting the target feature vector into structural data through an output layer of the graph neural network to obtain the intermediate recording information.

In one embodiment, the apparatus further comprises:

The analysis module is used for analyzing the intermediate form information to acquire the main risk type and the insured person characteristics;

The first screening module is used for screening a first additional risk set related to the main risk type according to the main risk type based on the insurance knowledge graph;

The second screening module is used for screening from the first additional risk set according to the characteristics of the insured person to obtain a second additional risk set;

The pushing module is used for sorting the second additional risk sets according to the monetary factor of each additional risk in the second additional risk sets, selecting the additional risk set with the preset rank from the sorted second additional risk sets as a target additional risk set, and pushing the target additional risk set.

And the error information error correction module 304 is configured to correct the intermediate recording information according to the insurance knowledge graph to obtain error information, and convert the intermediate recording information by using a sequence-to-sequence model according to the error information to generate target recording information.

In one embodiment, the apparatus further comprises:

the dictionary correction module is used for correcting the middle record information through a dictionary to obtain the middle record information subjected to dictionary correction;

The model error correction module is used for calculating the intermediate form information subjected to dictionary error correction through a random forest model to obtain the probability of an error statement;

And the marking module is used for marking the error statement to obtain the error information when the probability of the error statement exceeds a preset error threshold value.

In one embodiment, the error information correction module includes:

The information identification sub-module is used for identifying the error type and the position of the error information in the intermediate record information according to the type mark of the error information;

The coding submodule is used for coding the error information through an encoder from the sequence to the sequence model to obtain error information codes;

And the decoding sub-module is used for acquiring context information according to the position of the error information in the middle record information, decoding the error information through the sequence-to-sequence decoder according to the context information and the error type, and generating the target record information.

In the embodiment, a basis can be provided for error correction of subsequent recording information through an insurance knowledge graph constructed according to a preset insurance database, so that the accuracy of recording the insurance policy is indirectly improved;

In order to solve the technical problems, the embodiment of the application also provides equipment (computer equipment). Referring specifically to fig. 4, fig. 4 is a basic structural block diagram of a computer device according to the present embodiment.

The computer device 4 comprises a memory 41, a processor 42, a network interface 43 communicatively connected to each other via a system bus. It is noted that only a computer device 4 having a memory 41, a processor 42, a network interface 43 is shown in the figures, but it is understood that not all illustrated components are required to be implemented and that more or fewer components may be implemented instead. It will be appreciated by those skilled in the art that the computer device herein is a device capable of automatically performing numerical calculation and/or information processing according to a preset or stored instruction, and its hardware includes, but is not limited to, a microprocessor, an Application SPECIFIC INTEGRATED Circuit (ASIC), a Programmable gate array (Field-Programmable GATE ARRAY, FPGA), a digital Processor (DIGITAL SIGNAL Processor, DSP), an embedded device, and the like.

The computer equipment can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing equipment. The computer equipment can perform man-machine interaction with a user through a keyboard, a mouse, a remote controller, a touch pad or voice control equipment and the like.

The memory 41 includes at least one type of readable storage medium including flash memory, hard disk, multimedia card, card memory (e.g., SD or DX memory, etc.), random Access Memory (RAM), static Random Access Memory (SRAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), programmable Read Only Memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the storage 41 may be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (SMART MEDIA CARD, SMC), a Secure Digital (SD) card, a flash memory card (FLASH CARD) or the like, which are provided on the computer device 4. Of course, the memory 41 may also comprise both an internal memory unit of the computer device 4 and an external memory device. In this embodiment, the memory 41 is typically used to store an operating system and various application software installed on the computer device 4, such as computer readable instructions of a policy entry method. Further, the memory 41 may be used to temporarily store various types of data that have been output or are to be output.

The processor 42 may be a central processing unit (Central Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is configured to execute computer readable instructions stored in the memory 41 or process data, such as computer readable instructions for executing the policy entry method.

The network interface 43 may comprise a wireless network interface or a wired network interface, which network interface 43 is typically used for establishing a communication connection between the computer device 4 and other electronic devices.

In the implementation process of the electronic equipment, the basis can be provided for the error correction of the follow-up recording information through the insurance knowledge graph constructed according to the preset insurance database, and the accuracy of recording the insurance policy is indirectly improved;

The present application also provides another embodiment, namely, a storage medium (computer-readable storage medium) storing computer-readable instructions executable by at least one processor to cause the at least one processor to perform the steps of the policy entry method as described above.

In the implementation process of the computer readable storage medium, a basis can be provided for error correction of the follow-up recording information through an insurance knowledge graph constructed according to a preset insurance database, and the accuracy of recording the insurance information is indirectly improved;

The non-native company software tools or components present in the embodiments of the present application are presented by way of example only and are not representative of actual use.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present application.

It is apparent that the above-described embodiments are only some embodiments of the present application, but not all embodiments, and the preferred embodiments of the present application are shown in the drawings, which do not limit the scope of the patent claims. This application may be embodied in many different forms, but rather, embodiments are provided in order to provide a thorough and complete understanding of the present disclosure. Although the application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing description, or equivalents may be substituted for elements thereof. All equivalent structures made by the content of the specification and the drawings of the application are directly or indirectly applied to other related technical fields, and are also within the scope of the application.

Claims

1. A policy entry method, the method comprising:

2. The policy entry method of claim 1, wherein said constructing an insurance knowledge graph from said preset insurance database comprises:

3. The policy entry method of claim 1, wherein said converting said initial transcript information into text-to-graphic structure information comprises:

4. The policy entry method of claim 1, wherein the processing the text-map structural information using a map neural network to generate intermediate form information includes:

5. The policy entry method of claim 1, wherein after said processing said text-map structural information with said map neural network to generate intermediate form information, the method further comprises:

6. The policy entry method of claim 1, wherein prior to said error correcting said intermediate transcript information according to said insurance knowledge graph, the method further comprises:

7. The policy entry method of claim 1, wherein the converting the intermediate form information using a sequence-to-sequence model according to the error information to generate target form information includes:

8. A policy entry device, the device comprising:

9. A computer device, the computer device comprising:

at least one processor, and

A memory communicatively coupled to the at least one processor, wherein,

The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the policy entry method of any one of claims 1 to 7.

10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the policy entry method according to any one of claims 1 to 7.