A data correction method, device, computer equipment and storage medium

By building a family relationship network in the insurance policy business system and using the Transformers module to perform data correction, the problems of low accuracy and availability in traditional data correction methods are solved, and more efficient data correction results are achieved.

CN114911952BActive Publication Date: 2024-11-26CHINA PING AN LIFE INSURANCE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210574891.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-05-24
Publication Date
2024-11-26
Estimated Expiration
2042-05-24

AI Technical Summary

Technical Problem

Traditional data correction methods are filled based on the continuity or discrete characteristics of the data, which cannot reflect the true characteristics of the data, resulting in low accuracy and low availability of the corrected data.

Method used

By obtaining historical policy data in the policy business system, performing entity extraction and integration, building a family relationship network, using the 12-layer Transformers Encoder module for summary encoding and decoding, and correcting the data to be corrected.

Benefits of technology

Improve the accuracy and availability of data corrections, ensure that the corrected data has authenticity and availability, and solve the problems of low accuracy and low availability in traditional methods.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114911952B_ABST
    Figure CN114911952B_ABST
Patent Text Reader

Abstract

The embodiments of the present application belong to the field of data processing technology in artificial intelligence, and relate to a data correction method, device, computer equipment and storage medium based on a family relationship network. When it is necessary to correct the data of a certain user, the present application obtains the historical policy data corresponding to the user in the policy business system, extracts the core entity data of the historical policy data, and then constructs a family relationship network closely related to the user based on the core entity data. After obtaining the family relationship network of the user, the data of the user is corrected based on the family relationship network. Since the data in the policy business system itself has authenticity and availability, the constructed family relationship network also has authenticity and availability, thereby ensuring the effectiveness of subsequent data corrections and effectively solving the problems of low accuracy and low availability of traditional data correction methods.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of data processing technology in artificial intelligence, and in particular to a data correction method, device, computer equipment and storage medium based on a family relationship network. Background Art

[0002] With the rapid development of the Internet, the product marketing industry is increasingly dependent on user data information. The quality of user data directly affects the results of applying the data for marketing. User data has become the primary basis for tapping the potential value of users. However, due to various reasons in the early stage, the existing user data has problems such as errors or omissions in the data information, which directly affects the subsequent prediction results, marketing strategies, etc. Therefore, how to correct the existing user data has become a crucial issue.

[0003] There is a data correction method that uses value filling to correct data. Specifically, continuous data is usually filled with mean, mode, median, zero value, etc.; discrete data is usually filled with specified values ​​in combination with the business, so as to achieve the purpose of data correction.

[0004] However, the applicant found that the traditional data correction method is based on the continuity or discrete characteristics between the data, which cannot reflect the real data characteristics, resulting in low accuracy and low usability of the corrected data. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a data correction method, device, computer equipment and storage medium based on a family relationship network to solve the problems of low data accuracy and low usability in traditional data correction methods.

[0006] In order to solve the above technical problems, the embodiment of the present application provides a data correction method based on a family relationship network, which adopts the following technical solution:

[0007] receiving data to be corrected carrying a user identifier to be corrected;

[0008] Calling a business system, and obtaining historical policy data corresponding to the user identifier to be corrected in the business system;

[0009] Performing entity extraction operation on the historical insurance policy data to obtain entity text data and entity type data;

[0010] Performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data;

[0011] Performing a vector conversion operation on the historical policy data to obtain policy vector data;

[0012] The policy vector data and the entity fusion data are respectively input into the language representation model for summary encoding operation to obtain summary encoded data, wherein the language representation model is composed of 12 layers of Transformers Encoder modules, and a word-entity cross attention layer is set between the multi-head attention layer and the feedforward neural network layer of the Encoder module;

[0013] Decoding the summary coded data to obtain a target policy summary;

[0014] Constructing a user relationship network corresponding to the user identifier to be revised according to the target policy summary;

[0015] A data correction operation is performed on the data to be corrected according to the user relationship network.

[0016] In order to solve the above technical problems, the embodiment of the present application further provides a data correction device based on a family relationship network, which adopts the following technical solution:

[0017] The to-be-corrected data acquisition module is used to receive the to-be-corrected data carrying the to-be-corrected user identifier;

[0018] A historical data acquisition module, used for calling a business system and acquiring historical policy data corresponding to the user identifier to be corrected in the business system;

[0019] An entity extraction module, used to perform entity extraction operations on the historical policy data to obtain entity text data and entity type data;

[0020] A fusion module, used for performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data;

[0021] A vector conversion module, used for performing a vector conversion operation on the historical policy data to obtain policy vector data;

[0022] A summary encoding module, used for inputting the policy vector data and the entity fusion data into a language representation model for summary encoding operation to obtain summary encoding data, wherein the language representation model is composed of 12 layers of Transformers Encoder modules, and a word-entity cross attention layer is provided between the multi-head attention layer and the feedforward neural network layer of the Encoder module;

[0023] A decoding module, used for decoding the summary coded data to obtain a target policy summary;

[0024] A relationship network construction module, used to construct a user relationship network corresponding to the user identifier to be revised according to the target policy summary;

[0025] A data correction module is used to perform a data correction operation on the data to be corrected according to the user relationship network.

[0026] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0027] The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the data correction method based on the family relationship network as described above when executing the computer-readable instructions.

[0028] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0029] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the data correction method based on the family relationship network as described above are implemented.

[0030] The present application provides a data correction method based on a family relationship network, including: receiving data to be corrected carrying a user identifier to be corrected; calling a business system, and obtaining historical insurance policy data corresponding to the user identifier to be corrected in the business system; performing an entity extraction operation on the historical insurance policy data to obtain entity text data and entity type data; performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data; performing a vector conversion operation on the historical insurance policy data to obtain insurance policy vector data; respectively inputting the insurance policy vector data and the entity fusion data into a language representation model for summary encoding operation to obtain summary encoding data, wherein the language representation model is composed of an Encoder module of 12 layers of Transformers, and a word-entity cross-attention layer is provided between the multi-head attention layer and the feedforward neural network layer of the Encoder module; performing a decoding operation on the summary encoding data to obtain a target insurance policy summary; constructing a user relationship network corresponding to the user identifier to be corrected according to the target insurance policy summary; and performing a data correction operation on the data to be corrected according to the user relationship network. Compared with the prior art, the present application obtains historical policy data corresponding to the user in the policy business system, extracts core entity data of the historical policy data, and then constructs a family relationship network closely related to the user based on the core entity data. After obtaining the family relationship network of the user, the user's data is corrected based on the family relationship network. Since the data in the policy business system itself has authenticity and availability, the constructed family relationship network also has authenticity and availability, thereby ensuring the effectiveness of subsequent data corrections and effectively solving the problems of low accuracy and low availability of traditional data correction methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0033] Figure 2 is a flowchart of the implementation of the data correction method based on the family relationship network provided in the first embodiment of the present application;

[0034] Figure 3 It is a specific schematic diagram of the insurance policy provided in Example 1 of the present application;

[0035] Figure 4This is a specific schematic diagram of the family relationship diagram provided in Example 1 of the present application;

[0036] Figure 5 yes Figure 2 A flowchart of a specific implementation of step S04;

[0037] Figure 6 This is a specific schematic diagram of a home network relationship provided in Example 1 of the present application;

[0038] Figure 7 yes Figure 2 A flowchart of another specific implementation of step S204;

[0039] Figure 8 is another specific schematic diagram of the user data to be corrected provided in the first embodiment of the present application;

[0040] Fig. 9 This is another specific schematic diagram of the home network relationship provided in the first embodiment of the present application;

[0041] Fig.10 This is another specific schematic diagram of the home network relationship provided in the first embodiment of the present application;

[0042] Fig.11 yes Figure 2 A flowchart of a specific implementation of step S203;

[0043] Fig.12 yes Fig.11 A flowchart of a specific implementation of step S1101;

[0044] Fig.13 yes Fig.11 A flowchart of a specific implementation of step S1104;

[0045] Fig.14 yes Fig.11 A flowchart of a specific implementation of step S1105;

[0046] Fig.15 is a structural diagram of a data correction device based on a family relationship network provided in Embodiment 2 of the present application;

[0047] Fig.16 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0049] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0050] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0051] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0052] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0053] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV), laptop computers, desktop computers, etc.

[0054] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0055] It should be noted that the data correction method based on the family relationship network provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the data correction device based on the family relationship network is generally set in the server / terminal device.

[0056] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0057] Embodiment 1

[0058] Continue to refer Figure 2 , shows an implementation flow chart of the data correction method based on the family relationship network provided in Example 1 of the present application. For the sake of convenience, only the part related to the present application is shown.

[0059] The above-mentioned data correction method based on the family relationship network includes the following steps:

[0060] Step S201: receiving data to be corrected carrying a user identifier to be corrected.

[0061] Step S202: calling the business system to obtain the historical policy data corresponding to the user ID to be corrected in the business system.

[0062] Step S203: Perform entity extraction operations on the historical policy data to obtain entity text data and entity type data.

[0063] Step S204: performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data.

[0064] In the embodiment of the present application, the original text to be extracted is ,in For the document First, use text tools (such as Hanlp, etc.) to extract entities from the original text (called entity extraction), and each entity has an entity type (name, place name, etc.), thus forming an entity sequence and entity type sequence .in Represents the sequence entities, Similarly, the EntityEmbedding and TypeEmbedding layers are used to map the two to fixed vectors of the same dimension, generating vector sequences respectively. and Considering the importance of the entity and the impact of the entity length, this application needs to and To merge.

[0065] Step S205: Perform vector conversion operation on the historical policy data to obtain policy vector data.

[0066] In the embodiment of the present application, the original text data D is mapped to a vector of the same dimension through the WordEmbedding layer to obtain a matrix .

[0067] Step S206: Input the policy vector data and entity fusion data into the language representation model for summary encoding operation to obtain summary encoded data, wherein the language representation model is composed of 12 layers of Transformers' Encoder modules stacked together, and a word-entity cross-attention layer is set between the multi-head attention layer and the feedforward neural network layer of the Encoder module.

[0068] Step S207: Decode the summary encoded data to obtain the target policy summary.

[0069] Step S208: constructing a user relationship network corresponding to the user identifier to be revised according to the target policy summary.

[0070] In the embodiment of the present application, the user relationship network is mainly limited to the following relationships: couples, parents, children, grandparents, and grandchildren. In combination with the actual situation, starting from the data of the insurance policy, the user's relationship network data is improved.

[0071] In the embodiment of the present application, the family relationship network identification workflow based on the insurance policy is as follows:

[0072] (1) Based on the existing insurance policy data, determine the policyholder, the insured, and the beneficiary, and select the relationship between the three as: husband and wife, parents, children, grandparents, and grandchildren.

[0073] (2) From a business perspective and actual data, determine the primary key to identify a person. Here, a person is defined by the customer number (i.e., one customer number is one person). A person may have multiple customer numbers, and information can be used to assist in identifying one customer number in the final result.

[0074] (3) Construct a network using the extracted relationship pairs: policyholder-insured, insured-beneficiary.

[0075] In practical applications, see Figure 3 and Figure 4 , Figure 3 In Policy 1 and Policy 2, Customer A and Customer C are associated with each other through Customer B. In Policy 1 and Policy 3, Customer B and Customer D are associated with each other through Customer A. Finally, Customer A, Customer D and all four of them are associated with each other. And through age and gender, the relationship between them can be obtained as follows: Figure 4 As shown in the family relationship diagram, we finally get: the relationship between 4 people in a family, customer B is the father, customer C is the mother, customer A is the son, and customer D is the daughter-in-law.

[0076] In the embodiment of the present application, by expansion, the insurance relationship between these persons and who is the insured can be found out, so as to determine who is the dominant person in the family in purchasing insurance policies, and marketing can be conducted to the dominant person. Secondly, the protection gap can be analyzed in a targeted manner based on the composition of the insurance policies in the family, age, etc., and the insurance product portfolio can be optimized.

[0077] Step S209: performing a data correction operation on the data to be corrected according to the user relationship network.

[0078] In the embodiment of the present application, after the home network is constructed, some data quality problems that often occur among family members, such as missing or incorrect marital status (error types include value errors, or corresponding to multiple marital statuses), multiple spouses due to divorce and marriage, inability to distinguish the current spouse, missing or incorrect age, missing relationships between certain members, etc., can be corrected more conveniently by bringing family members together through the home network.

[0079] In an embodiment of the present application, a data correction method based on a family relationship network is provided, including: receiving data to be corrected carrying a user identifier to be corrected; calling a business system, and obtaining historical policy data corresponding to the user identifier to be corrected in the business system; performing an entity extraction operation on the historical policy data to obtain entity text data and entity type data; performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data; performing a vector conversion operation on the historical policy data to obtain policy vector data; respectively inputting the policy vector data and the entity fusion data into a language representation model for summary encoding operation to obtain summary encoding data, wherein the language representation model is composed of an Encoder module of 12 layers of Transformers, and a word-entity cross-attention layer is provided between the multi-head attention layer and the feedforward neural network layer of the Encoder module; performing a decoding operation on the summary encoding data to obtain a target policy summary; constructing a user relationship network corresponding to the user identifier to be corrected according to the target policy summary; and performing a data correction operation on the data to be corrected according to the user relationship network. Compared with the prior art, the present application obtains historical policy data corresponding to the user in the policy business system, extracts core entity data of the historical policy data, and then constructs a family relationship network closely related to the user based on the core entity data. After obtaining the family relationship network of the user, the user's data is corrected based on the family relationship network. Since the data in the policy business system itself has authenticity and availability, the constructed family relationship network also has authenticity and availability, thereby ensuring the effectiveness of subsequent data corrections and effectively solving the problems of low accuracy and low availability of traditional data correction methods.

[0080] Continue reading Figure 5 , showing Figure 2 A flowchart of a specific implementation of step S209 is shown in FIG. 1 . For ease of explanation, only the part related to the present application is shown.

[0081] In some optional implementations of this embodiment, step S209 specifically includes:

[0082] Step S501: if the marital status of the data to be corrected is missing, then obtain the child user data corresponding to the user ID to be corrected;

[0083] Step S502: determining whether there is parent user data corresponding to the child user data in the user relationship network;

[0084] Step S503: If the parent user data exists, the marital status is corrected to married, and a husband-wife association relationship is added to the parent user data;

[0085] Step S504: If there is only data to be corrected, the marital status is corrected to unmarried.

[0086] In the embodiment of the present application, when the marital status is missing, for example, a family member has a missing marital status, but the family network relationship is Figure 6 When shown, Zhang San's marital status is missing, but according to Figure 6 From Zhang Xiaosan, we can know that Zhang San has a son, Zhang Xiaosan, and Zhang Xiaosan has a mother, Li Si. Zhang San and Li Si are close in age, and Li Si is married. Through this relationship, we can infer with a high probability that Zhang San is married, and we can conclude that Zhang San and Li Si are husband and wife.

[0087] Continue reading Figure 7 , showing Figure 2 A flowchart of another specific implementation of step S209 is shown, for the sake of convenience of explanation, only the part related to the present application is shown.

[0088] In some optional implementations of this embodiment, step S209 specifically includes:

[0089] Step S701: if the data to be corrected has several unique associations, obtain all associated user data that have a unique association with the user identifier to be corrected;

[0090] Step S702: acquiring the associated user data with the earliest insurance policy underwriting time from the associated user data;

[0091] Step S703: only establish a unique association relationship between the user ID to be modified and the associated user data with the earliest insurance policy underwriting time.

[0092] In the embodiments of the present application, Figure 8 As shown, Zhang San has two spouses in the insurance policy, but judging from the latest insurance time of the respective policies, Wang Wu’s latest insurance time is later than Li Si’s, so it is inferred that Wang Wu is Zhang San’s current spouse, and Li Si is most likely Zhang San’s former spouse.

[0093] In some optional implementations of this embodiment, see Fig. 9 ,from Fig. 9 From the family composition of the family, we can deduce that Zhang Xiaoshuai and Zhang Xiaomei are siblings, and the relationship between any two other members can be obtained.

[0094] In some optional implementations of this embodiment, see Fig.10 ,from Fig.10 In the family relationship, Zhang San’s gender is missing, but through his father-son relationship with Zhang Xiaosan, we can get that Zhang San’s gender is male.

[0095] Continue reading Fig.11 , showing Figure 2 A flowchart of a specific implementation of step S208 is shown in FIG. 1 . For ease of explanation, only the part related to the present application is shown.

[0096] In some optional implementations of this embodiment, the target policy summary includes at least one of the information of the policyholder, the information of the insured, and the information of the beneficiary. Step S208 specifically includes:

[0097] Step S1101: extracting the policyholder information, the insured person information and the beneficiary information of each target policy summary, as well as the first initial family relationship of the policyholder information, the insured person information and the beneficiary information in each target policy summary.

[0098] In an embodiment of the present application, when extracting information from a target policy summary, the same method as extracting information from at least one policy data is used for extraction. The difference here is that it is necessary to extract all user information in the target policy summary and the relationship between users, and based on this relationship, the family relationship of each target policy summary is constructed to obtain the first initial family relationship.

[0099] In an embodiment of the present application, before extracting information, it also includes identifying the data format of each historical insurance policy. If it is a picture, image recognition technology is used to convert the text to form insurance policies in formats such as PDF, tables, and WORD, and then customer information is extracted by identifying personal information fields in the insurance policy.

[0100] Step S1102: Use the policyholder information, the insured person information and the beneficiary information in each target policy summary as matching dimensions, and perform pairwise matching with other target policy summaries to obtain matching results.

[0101] In an embodiment of the present application, first, one target policy summary is selected from multiple target policy summaries as a matching object, and then the matching object is matched with all remaining target policy summaries in pairs to obtain a matching result.

[0102] In an embodiment of the present application, in the process of pairwise matching, it is necessary to select one of the policyholder information, insured person information and beneficiary information in the matching object as a matching criterion, and then use the matching criterion to select the target policy summaries with the same matching criterion from all the remaining target policy summaries to obtain the matching status of the matching object and each target policy summary.

[0103] Step S1103: According to the matching results, the target policy summaries with the same matching dimension are screened out to obtain at least two first policy data sets.

[0104] In an embodiment of the present application, the matching result is the degree of matching between the target policy summaries, and is output in the form of a specific matched result table. That is, after the above matching is completed, the matching objects are respectively matched with the remaining target policy summaries and the policy numbers to generate a relationship table. When filtering target policy summaries with the same dimension here, the relationship table can be directly identified.

[0105] Step S1104: extracting the first initial family relationship of each target policy summary in each first policy data set, and constructing a second initial family relationship based on the first initial family relationship.

[0106] In the embodiment of the present application, each first insurance policy data set is used as a processing unit, and the first initial family relationship of each target insurance policy summary in the first insurance policy data set is extracted, and the second initial family relationship is constructed based on the first initial family relationship, which specifically includes:

[0107] 1) Acquire corresponding customer information based on the first initial family relationship, where the customer information is the information of the policyholder, the insured, and the beneficiary in the corresponding target insurance policy summary;

[0108] 2) Using customer information as element points, construct an element matrix;

[0109] 3) Adjust the position of each element point in the element matrix according to the first initial family relationship to obtain a family matrix;

[0110] 4) Convert the family matrix into the second initial family relationship.

[0111] Step S1105: construct a family relationship network of all target policy summaries based on all second initial family relationships to obtain a user relationship network.

[0112] In the embodiment of the present application, the construction here is carried out by merging user information. Specifically, two are selected from all the second initial family relationships as the first merged object group, and the nodes in one second initial family relationship in the first merged object group are matched with the nodes of another second initial family relationship. If the match is the same, the two nodes are merged until all the nodes in the first merged object group are matched, and a new second initial family relationship is output, and the new second initial family relationship is matched one by one with the nodes of the next second initial family relationship until all the second initial family relationships are matched, and a family relationship network is obtained.

[0113] Continue reading Fig.12 , showing Fig.11 A flowchart of a specific implementation of step S1101 is shown in FIG. 1 . For ease of explanation, only the part related to the present application is shown.

[0114] In some optional implementations of this embodiment, step S1101 specifically includes:

[0115] Step S1201: Encode each target policy summary in the policy data set to obtain a corresponding data storage code.

[0116] In the embodiment of the present application, the coding process is specifically to extract the corresponding code from the background database of the business system by extracting the policy number, or directly converting the target policy summary into code. The code conversion is achieved through a preset code conversion tool. Specifically, the type of policy is first determined, and the corresponding policy form template is selected based on the type of policy. Then, text extraction technology is used to extract text information and the position of the text in the policy from the target policy summary. The position is matched with each position in the policy form template. Based on the matching result, the corresponding text information is filled into the policy form template to generate the policy. Then, the policy is converted into a code file to obtain the data storage code.

[0117] Step S1202: Identify the user information tag in the data storage code, and extract the content corresponding to the user information tag to obtain the policyholder information, the insured person information, the beneficiary information, and the relationship attributes among the three recorded in each target policy summary.

[0118] In an embodiment of the present application, the data type of the user information is determined to obtain a user information tag, and based on the user information tag, the string of the corresponding data type and the name of the string in the data storage code are queried, and the actual content of the string is determined based on the name. For example, a string named policyholder information is the name of the customer, and a string named relationship attribute is a family relationship.

[0119] Step S1203: construct a relationship tree diagram based on the relationship attributes of the policyholder information, the insured person information and the beneficiary information of each historical data policy to obtain the first initial family relationship.

[0120] Continue reading Fig.13 , showing Fig.11 A flowchart of a specific implementation of step S1104 is shown in FIG. 1 , and for ease of explanation, only the part related to the present application is shown.

[0121] In some optional implementations of this embodiment, step S1104 specifically includes:

[0122] Step S1301: Obtain a relationship tree diagram corresponding to the first initial family relationship of each target policy summary in each first policy data set.

[0123] Step S1302: Match the leaf nodes and trunk nodes in each relationship tree diagram.

[0124] Step S1303: If the matching result is that the leaf nodes are the same, the same leaf nodes in all corresponding relationship tree diagrams are associated to obtain a relationship tree diagram group.

[0125] Step S1304: If the matching result is that the trunk nodes are the same, the same trunk nodes in all corresponding relationship tree diagrams are associated to obtain a relationship tree diagram group.

[0126] Step S1305: until there are no identical and separate leaf nodes or trunk nodes among all the relationship tree diagrams, the relationship tree diagram group is output to obtain the second initial family relationship.

[0127] In the embodiment of the present application, when matching nodes, specifically, leaf nodes are matched with leaf nodes and trunk nodes in other relationship tree diagrams respectively, and as long as the matches are consistent, it is determined to be a merge relationship.

[0128] In the embodiment of the present application, the trunk node is matched with the leaf nodes and trunk nodes in other relationship tree diagrams, and as long as the matches are consistent, it is determined to be a merge relationship.

[0129] In an embodiment of the present application, the merging process is as follows: first, based on the result of matching the trunk node, after matching the trunk node, the trunk nodes in the two relationship tree diagrams are merged, and based on the trunk node, the leaf nodes in the two relationship tree diagrams are added to the trunk node upwards and downwards respectively to obtain a second initial family relationship.

[0130] Continue reading Fig.14 , showing Fig.11 A flowchart of a specific implementation of step S1105 is shown in which, for ease of explanation, only the part related to the present application is shown.

[0131] In some optional implementations of this embodiment, step S1105 specifically includes:

[0132] Step S1401: Match leaf nodes and trunk nodes in all second initial family relationships in pairs, and calculate the node intersection of all second initial family relationships;

[0133] Step S1402: associate the corresponding second initial family relationship nodes according to the node intersection to obtain the family relationship network of all historical insurance policies.

[0134] In the embodiment of the present application, after calculating the node intersection, it also includes calculating the intimacy coefficient between the leaf nodes and the trunk nodes.

[0135] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. The aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0136] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0137] Embodiment 2

[0138] Further references Fig.15 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a data correction device based on a family relationship network. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0139] like Fig.15 As shown, the data correction device 200 based on the family relationship network of this embodiment includes: a data acquisition module 210 to be corrected, a historical data acquisition module 220, an entity extraction module 230, a fusion module 240, a vector conversion module 250, a summary encoding module 260, a decoding module 270, a relationship network construction module 280 and a data correction module 290. Among them:

[0140] The to-be-corrected data acquisition module 210 is used to receive the to-be-corrected data carrying the to-be-corrected user identifier;

[0141] The historical data acquisition module 220 is used to call the business system and acquire the historical policy data corresponding to the user identifier to be corrected in the business system;

[0142] An entity extraction module 230 is used to perform entity extraction operations on the historical policy data to obtain entity text data and entity type data;

[0143] A fusion module 240 is used to perform a fusion operation on the entity text data and the entity type data to obtain entity fusion data;

[0144] A vector conversion module 250 is used to perform a vector conversion operation on the historical insurance policy data to obtain insurance policy vector data;

[0145] A summary encoding module 260 is used to input the policy vector data and the entity fusion data into a language representation model for summary encoding to obtain summary encoding data, wherein the language representation model is composed of 12 layers of Transformers Encoder modules, and a word-entity cross attention layer is provided between the multi-head attention layer and the feedforward neural network layer of the Encoder module;

[0146] A decoding module 270, configured to decode the summary coded data to obtain a target policy summary;

[0147] A relationship network construction module 280, configured to construct a user relationship network corresponding to the user identifier to be revised according to the target policy summary;

[0148] The data correction module 290 is used to perform a data correction operation on the data to be corrected according to the user relationship network.

[0149] In the embodiment of the present application, the user relationship network is mainly limited to the following relationships: couples, parents, children, grandparents, and grandchildren. In combination with the actual situation, starting from the data of the insurance policy, the user's relationship network data is improved.

[0150] In the embodiment of the present application, the family relationship network identification workflow based on the insurance policy is as follows:

[0151] (1) Based on the existing insurance policy data, determine the policyholder, the insured, and the beneficiary, and select the relationship between the three as: husband and wife, parents, children, grandparents, and grandchildren.

[0152] (2) From a business perspective and actual data, determine the primary key to identify a person. Here, a person is defined by the customer number (i.e., one customer number is one person). A person may have multiple customer numbers, and information can be used to assist in identifying one customer number in the final result.

[0153] (3) Construct a network using the extracted relationship pairs: policyholder-insured, insured-beneficiary.

[0154] In practical applications, see Figure 3 and Figure 4 , Figure 3 In Policy 1 and Policy 2, Customer A and Customer C are associated with each other through Customer B. In Policy 1 and Policy 3, Customer B and Customer D are associated with each other through Customer A. Finally, Customer A, Customer D and all four of them are associated with each other. And through age and gender, the relationship between them can be obtained as follows: Figure 4 As shown in the family relationship diagram, we finally get: the relationship between 4 people in a family, customer B is the father, customer C is the mother, customer A is the son, and customer D is the daughter-in-law.

[0155] In the embodiment of the present application, by expansion, the insurance relationship between these persons and who is the insured can be found out, so as to determine who is the dominant person in the family in purchasing insurance policies, and marketing can be conducted to the dominant person. Secondly, the protection gap can be analyzed in a targeted manner based on the composition of the insurance policies in the family, age, etc., and the insurance product portfolio can be optimized.

[0156] In the embodiment of the present application, after the home network is constructed, some data quality problems that often occur among family members, such as missing or incorrect marital status (error types include value errors, or corresponding to multiple marital statuses), multiple spouses due to divorce and marriage, inability to distinguish the current spouse, missing or incorrect age, missing relationships between certain members, etc., can be corrected more conveniently by bringing family members together through the home network.

[0157] In an embodiment of the present application, a data correction device 200 based on a family relationship network is provided, including: a data acquisition module 210 for receiving data to be corrected carrying a user identifier to be corrected; a historical data acquisition module 220 for calling a business system to obtain historical insurance policy data corresponding to the user identifier to be corrected in the business system; an entity extraction module 230 for performing an entity extraction operation on the historical insurance policy data to obtain entity text data and entity type data; a fusion module 240 for performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data; a vector conversion module 250 for performing a vector conversion operation on the historical insurance policy data to obtain insurance policy vector data; and a summary encoding module 260, used to input the policy vector data and the entity fusion data into the language representation model for summary encoding operation to obtain summary encoded data, wherein the language representation model is composed of 12 layers of Transformers Encoder modules, and a word-entity cross-attention layer is provided between the multi-head attention layer and the feedforward neural network layer of the Encoder module; a decoding module 270, used to decode the summary encoded data to obtain a target policy summary; a relationship network construction module 280, used to construct a user relationship network corresponding to the user identifier to be corrected according to the target policy summary; a data correction module 290, used to perform a data correction operation on the data to be corrected according to the user relationship network. Compared with the prior art, the present application obtains historical policy data corresponding to the user in the policy business system, extracts core entity data of the historical policy data, and then constructs a family relationship network closely related to the user based on the core entity data. After obtaining the family relationship network of the user, the user's data is corrected based on the family relationship network. Since the data in the policy business system itself has authenticity and availability, the constructed family relationship network also has authenticity and availability, thereby ensuring the effectiveness of subsequent data corrections and effectively solving the problems of low accuracy and low availability of traditional data correction methods.

[0158] In some optional implementations of this embodiment, the data correction module 290 includes:

[0159] The child data acquisition submodule is used to acquire the child user data corresponding to the user identifier to be modified if the marital status of the data to be modified is missing;

[0160] The parent data judgment submodule is used to judge whether there is parent user data corresponding to the child user data in the user relationship network;

[0161] The first marital status correction submodule is used to correct the marital status to married if the parent user data exists, and add a husband-wife association relationship to the parent user data;

[0162] The second marital status correction submodule is used to correct the marital status to unmarried if there is only data to be corrected.

[0163] In some optional implementations of this embodiment, the data correction module 290 further includes:

[0164] The associated data acquisition submodule is used to acquire all associated user data that have a unique association relationship with the user identifier to be corrected if the data to be corrected has several unique association relationships;

[0165] The associated data screening submodule is used to obtain the associated user data with the earliest insurance policy underwriting time from the associated user data;

[0166] The association confirmation submodule is used to establish only a unique association corresponding to the user identification to be corrected and the associated user data with the earliest insurance policy underwriting time.

[0167] In some optional implementations of this embodiment, the relationship network construction module 280 includes:

[0168] A first initial relationship extraction submodule, used to extract the policyholder information, the insured information and the beneficiary information of each target policy summary, and the first initial family relationship of the policyholder information, the insured information and the beneficiary information in each target policy summary;

[0169] The matching submodule is used to use the policyholder information, the insured information and the beneficiary information in each target policy summary as matching dimensions, and to perform pairwise matching with other target policy summaries to obtain matching results;

[0170] A screening submodule, configured to screen out target insurance policy summaries with the same matching dimension according to the matching results, and obtain at least two first insurance policy data sets;

[0171] A second initial relationship construction submodule is used to extract the first initial family relationship of each target insurance policy summary in each first insurance policy data set, and construct a second initial family relationship based on the first initial family relationship;

[0172] The relationship network construction submodule is used to construct a family relationship network of all target insurance policy summaries based on all second initial family relationships to obtain a user relationship network.

[0173] In some optional implementations of this embodiment, the first initial relationship extraction submodule includes:

[0174] A coding unit, used for coding each target policy summary in the policy data set to obtain a corresponding data storage code;

[0175] A relationship attribute extraction unit is used to identify the user information tag in the data storage code and extract the content corresponding to the user information tag to obtain the policyholder information, the insured person information, the beneficiary information, and the relationship attributes of the three recorded in each target policy summary;

[0176] The tree diagram construction unit is used to construct a relationship tree diagram based on the relationship attributes of the policyholder information, the insured person information and the beneficiary information of each historical data insurance policy to obtain the first initial family relationship.

[0177] In some optional implementations of this embodiment, the second initial relationship building submodule includes:

[0178] A tree diagram acquisition unit, used to acquire a relationship tree diagram corresponding to the first initial family relationship of each target insurance policy summary in each first insurance policy data set;

[0179] A first node matching unit, used for matching leaf nodes and trunk nodes in each relationship tree diagram with each other;

[0180] The leaf node association unit is used to associate the same leaf nodes in all corresponding relationship tree diagrams to obtain a relationship tree diagram group if the matching result shows that the leaf nodes are the same;

[0181] A trunk node associating unit, for associating the same trunk nodes in all corresponding relationship tree diagrams to obtain a relationship tree diagram group if the matching result shows that the trunk nodes are the same;

[0182] The second initial relationship construction unit is used to output the relationship tree diagram group until there are no identical and separated leaf nodes or trunk nodes between all the relationship tree diagrams to obtain the second initial family relationship.

[0183] In some optional implementations of this embodiment, the relationship network construction submodule includes:

[0184] A second node matching unit is used to match leaf nodes and trunk nodes in all second initial family relationships in pairs, and calculate the node intersection of all second initial family relationships;

[0185] The relationship network construction unit is used to associate the corresponding second initial family relationship nodes according to the node intersection to obtain the family relationship network of all historical insurance policies.

[0186] To solve the above technical problems, the present application also provides a computer device. Fig.16 , Fig.16This is a basic structural block diagram of the computer device in this embodiment.

[0187] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 300 having components 310-330, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0188] The computer device may be a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with the user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0189] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, 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 memory 310 can be an internal storage unit of the computer device 300, such as a hard disk or memory of the computer device 300. In other embodiments, the memory 310 can also be an external storage device of the computer device 300, such as a plug-in hard disk equipped on the computer device 300, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. Of course, the memory 310 can also include both the internal storage unit of the computer device 300 and its external storage device. In this embodiment, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions of the data correction method based on the family relationship network, etc. In addition, the memory 310 can also be used to temporarily store various data that have been output or are to be output.

[0190] The processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 320 is generally used to control the overall operation of the computer device 300. In this embodiment, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as running the computer-readable instructions of the data correction method based on the family relationship network.

[0191] The network interface 330 may include a wireless network interface or a wired network interface. The network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0192] The computer device provided by the present application obtains historical policy data corresponding to the user in the policy business system, and constructs a family relationship network closely related to the user based on the historical policy data. After obtaining the family relationship network of the user, the user's data is corrected based on the family relationship network. Since the data in the policy business system itself has authenticity and availability, the constructed family relationship network also has authenticity and availability, thereby ensuring the effectiveness of subsequent data corrections and effectively solving the problems of low accuracy and low availability of traditional data correction methods.

[0193] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the data correction method based on the family relationship network as described above.

[0194] The computer-readable storage medium provided by the present application obtains historical policy data corresponding to the user in the policy business system, and constructs a family relationship network closely related to the user based on the historical policy data. After obtaining the family relationship network of the user, the data of the user is corrected according to the family relationship network. Since the data in the policy business system itself has authenticity and availability, the constructed family relationship network also has authenticity and availability, thereby ensuring the effectiveness of subsequent data corrections and effectively solving the problems of low accuracy and low availability of traditional data correction methods.

[0195] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0196] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A data correction method based on family relationship network, It is characterized in that The steps include: receiving data to be corrected carrying a user identifier to be corrected; Calling a business system, and obtaining historical policy data corresponding to the user identifier to be corrected in the business system; Performing entity extraction operation on the historical insurance policy data to obtain entity text data and entity type data; Performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data; Performing a vector conversion operation on the historical policy data to obtain policy vector data; The policy vector data and the entity fusion data are respectively input into the language representation model for summary encoding operation to obtain summary encoded data, wherein the language representation model is composed of 12 layers of Transformers Encoder modules, and a word-entity cross attention layer is set between the multi-head attention layer and the feedforward neural network layer of the Encoder module; Decoding the summary coded data to obtain a target policy summary; Constructing a user relationship network corresponding to the user identifier to be revised according to the target policy summary; A data correction operation is performed on the data to be corrected according to the user relationship network.

2. The data correction method based on the family relationship network according to claim 1, It is characterized in that The step of performing a data correction operation on the data to be corrected according to the user relationship network specifically includes the following steps: If the marital status of the data to be corrected is missing, obtaining the child user data corresponding to the user identifier to be corrected; Determining whether there is parent user data corresponding to the child user data in the user relationship network; If parent user data exists, the marital status is corrected to married, and a husband-wife association relationship is added to the parent user data; If only the data to be corrected exists, the marital status is corrected to unmarried status.

3. The data correction method based on the family relationship network according to claim 1, It is characterized in that The step of performing a data correction operation on the data to be corrected according to the user relationship network specifically includes the following steps: If the data to be corrected has several unique associations, obtaining all associated user data that have the unique association with the user identifier to be corrected; Acquire the associated user data with the earliest insurance policy underwriting time from the associated user data; Only the unique association relationship corresponding to the user identification to be corrected and the associated user data with the earliest insurance policy underwriting time is established.

4. The data correction method based on the family relationship network according to claim 1, It is characterized in that The target insurance policy summary includes at least one of the information of the policyholder, the information of the insured, and the information of the beneficiary. The step of constructing a user relationship network corresponding to the user identifier to be modified according to the target insurance policy summary specifically includes the following steps: Extracting the policyholder information, the insured person information and the beneficiary information of each of the target policy summaries, as well as the first initial family relationship of the policyholder information, the insured person information and the beneficiary information in each of the target policy summaries; The insured information, the insured person information and the beneficiary information in each target policy summary are used as matching dimensions respectively, and are matched with other target policy summaries in pairs to obtain matching results; According to the matching results, target insurance policy summaries with the same matching dimension are screened out to obtain at least two first insurance policy data sets; Extracting a first initial family relationship of each target insurance policy summary in each of the first insurance policy data sets, and constructing a second initial family relationship based on the first initial family relationship; A family relationship network of all target insurance policy summaries is constructed based on all the second initial family relationships to obtain the user relationship network.

5. The data correction method based on the family relationship network according to claim 4, It is characterized in that The step of extracting the policyholder information, the insured person information and the beneficiary information of each target policy summary, as well as the first initial family relationship of the policyholder information, the insured person information and the beneficiary information in each target policy summary, specifically comprises the following steps: Coding each target policy summary in the policy data set to obtain a corresponding data storage code; Identify the user information tag in the data storage code, extract the content corresponding to the user information tag, and obtain the policyholder information, the insured person information, the beneficiary information, and the relationship attributes between the three recorded in each target policy summary; Based on the relationship attributes, the policyholder information, the insured person information and the beneficiary information of each historical data insurance policy are used to construct a relationship tree diagram to obtain the first initial family relationship.

6. The data correction method based on the family relationship network according to claim 4, It is characterized in that The step of extracting the first initial family relationship of each target insurance policy summary in each of the first insurance policy data sets and constructing the second initial family relationship based on the first initial family relationship specifically includes the following steps: Obtaining a relationship tree diagram corresponding to a first initial family relationship of each target insurance policy summary in each of the first insurance policy data sets; Matching leaf nodes and trunk nodes in each of the relationship tree diagrams; If the matching result is that the leaf nodes are the same, then all the corresponding identical leaf nodes in the relationship tree diagrams are associated to obtain a relationship tree diagram group; If the matching result is that the trunk nodes are the same, then the same trunk nodes in all corresponding relationship tree diagrams are associated to obtain a relationship tree diagram group; Until there are no identical and separated leaf nodes or trunk nodes among all the relationship tree diagrams, the relationship tree diagram group is output to obtain the second initial family relationship.

7. The data correction method based on the family relationship network according to claim 4, It is characterized in that The step of constructing a family relationship network of all target insurance policy summaries based on all the second initial family relationships to obtain the user relationship network specifically includes the following steps: Matching leaf nodes and trunk nodes in all the second initial family relationships in pairs, and calculating the node intersection of all the second initial family relationships; The corresponding second initial family relationship is node-associated according to the node intersection to obtain a family relationship network of all historical insurance policies.

8. A data correction device based on a family relationship network, It is characterized in that include: The to-be-corrected data acquisition module is used to receive the to-be-corrected data carrying the to-be-corrected user identifier; A historical data acquisition module, used for calling a business system and acquiring historical policy data corresponding to the user identifier to be corrected in the business system; An entity extraction module, used to perform entity extraction operations on the historical policy data to obtain entity text data and entity type data; A fusion module, used for performing a fusion operation on the entity text data and the entity type data to obtain entity fusion data; A vector conversion module, used for performing a vector conversion operation on the historical policy data to obtain policy vector data; A summary encoding module, used for inputting the policy vector data and the entity fusion data into a language representation model for summary encoding operation to obtain summary encoding data, wherein the language representation model is composed of 12 layers of Transformers Encoder modules, and a word-entity cross attention layer is provided between the multi-head attention layer and the feedforward neural network layer of the Encoder module; A decoding module, used for decoding the summary coded data to obtain a target policy summary; A relationship network construction module, used to construct a user relationship network corresponding to the user identifier to be revised according to the target policy summary; A data correction module is used to perform a data correction operation on the data to be corrected according to the user relationship network.

9. A computer device, It is characterized in that It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the data correction method based on the family relationship network as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the data correction method based on a family relationship network as described in any one of claims 1 to 7 are implemented.