Insurance data analysis system, method and equipment

Abstract

The invention provides an insurance data analysis system, method and device, and relates to the technical field of data analysis, and the system can receive and understand a data query request of a user through a first data processing unit, and accurately capture the intention of the user based on an advanced natural language processing technology, thereby generating an accurate data query instruction. In addition, three built-in data analysis units of the system can work cooperatively, and deep mining and analysis are carried out on insurance data according to a data query instruction. According to the multi-unit cooperation mode, the efficiency and accuracy of data analysis are improved, comprehensive analysis results can be provided from multiple dimensions, and diversified requirements of users are met. And finally, the second data processing unit processes the first processing result again, so that the quality and readability of the output result are further optimized. The multi-level data processing flow ensures the accuracy and practicability of a final output result, and a clearer and more visual insurance data analysis result is provided for a user.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to an insurance data analysis system, method and device. Background Art

[0002] In the practical scenarios of insurance business analysis, the data analysis tasks faced by front-line business personnel are often highly complex and diverse, covering many indicators and specific application scenarios, and are accompanied by a series of modular and fragmented workflows such as data search, extraction, and analysis. Traditional data processing methods require business personnel to not only master SQL programming skills, but also have an in-depth understanding of metadata assets and excellent business data analysis capabilities. However, in actual operations, business personnel often need to cross multiple functional platforms to complete the entire data analysis process, which undoubtedly increases the complexity and time cost of operations. Specifically, the complexity of insurance data assets is reflected in the numerous fields, diverse calibers, wide application scenarios, and diversified data sources, which brings great challenges to business personnel in finding, confirming, understanding, and using data. In the data extraction stage, since SQL code needs to be written, this is undoubtedly an insurmountable obstacle for business personnel who do not have relevant skills. In addition, the insurance business analysis and data analysis fields have extremely strict requirements on data accuracy, especially when it comes to specific analysis scenarios such as institutional rankings, any slight error may lead to distortion of analysis results.

[0003] Therefore, there is an urgent need for an insurance data analysis method that can overcome the above-mentioned defects and improve the efficiency and accuracy of data analysis. Summary of the invention

[0004] In view of the above problems, the present application provides an insurance data analysis system, method and device, including the following contents:

[0005] In a first aspect, the present application provides an insurance data analysis system, the system comprising:

[0006] A first data processing unit, a second data processing unit, a first data analysis unit, a second data analysis unit, and a third data analysis unit;

[0007] The first data processing unit is used to receive a user's data query request, understand the user's query intention based on natural language processing technology, and generate a data query instruction; input the data query instruction into one or more of the corresponding first data analysis unit, second data analysis unit and third data analysis unit for processing to obtain a first processing result;

[0008] The second data processing unit is used to reprocess the data in the first processing result to obtain the final output result.

[0009] Optionally, the system further includes:

[0010] The metadata management unit is used to store and manage relevant data, knowledge, metadata information and analysis methods in the insurance field, and supports rapid data retrieval and intelligent question and answer.

[0011] Optionally, the first data analysis unit is specifically used for:

[0012] Vectorizing the data query instruction, and performing a search based on the vectorized data, or directly performing a keyword search based on the data query instruction;

[0013] The answer is queried from the metadata management unit by using the retrieval enhancement to generate RAG.

[0014] Optionally, the second data analysis unit is specifically used to:

[0015] Extracting keywords from the data query instruction;

[0016] Generate corresponding SQL query statements based on the extracted keywords or compose domain-specific language DSL for query;

[0017] Execute query operations in the database based on the generated query statements, and render the query results as visual components for display.

[0018] Optionally, the third data analysis unit is specifically used for:

[0019] Extracting parameters and instructions from the data query instruction using a large model;

[0020] Match the prepared insurance data analysis general API and execute the relevant API according to the extracted parameters and instructions;

[0021] After obtaining the query data results, the query data results are interpreted and summarized using the big model to form a conclusion report;

[0022] Return the conclusion report to the user.

[0023] Optionally, the metadata management unit further includes a knowledge segmentation module, a vectorization module and an index creation module, wherein:

[0024] The knowledge segmentation module is used to segment the acquired insurance domain knowledge;

[0025] The vectorization module is used to vectorize the knowledge after segmentation;

[0026] The index building module is used to build an index for the vectorized knowledge.

[0027] Optionally, the first data processing unit includes a context combination module, and the context combination module is used to determine whether the current user query request is associated with a previous query request, and adjust the content of the model input based on the context information.

[0028] Optionally, the second data processing unit includes an answer generation module, which is used to generate a final user answer based on the first processing result generated by the first data processing unit in combination with context information, and return it to the user.

[0029] In a second aspect, the present application provides an insurance data analysis method, the method comprising:

[0030] Obtain the user's data query request, understand the user's query intent based on natural language processing technology, and generate data query instructions;

[0031] Inputting the data query instruction into one or more of the first data analysis unit, the second data analysis unit and the third data analysis unit for processing to obtain a first processing result;

[0032] The data in the first processing result is processed again to obtain the final output result.

[0033] Optionally, the method further includes:

[0034] The relevant data, knowledge, metadata information and analysis methods in the insurance field are managed through a metadata management unit; the metadata management unit is used to perform rapid retrieval of data and intelligent question and answer.

[0035] Optionally, inputting the data query instruction into the first data analysis unit for processing includes:

[0036] Vectorizing the data query instruction, and performing a search based on the vectorized data, or directly performing a keyword search based on the data query instruction;

[0037] The answer is queried from the metadata management unit by using the retrieval enhancement to generate RAG.

[0038] Optionally, inputting the data query instruction into the second data analysis unit for processing includes:

[0039] Extracting keywords from the data query instruction;

[0040] Generate corresponding SQL query statements based on the extracted keywords or compose domain-specific language DSL for query;

[0041] Execute query operations in the database based on the generated query statements, and render the query results as visual components for display.

[0042] Optionally, inputting the data query instruction into a third data analysis unit for processing includes:

[0043] Extracting parameters and instructions from the data query instruction using a large model;

[0044] Match the prepared insurance data analysis general API and execute the relevant API according to the extracted parameters and instructions;

[0045] After obtaining the query data results, the query data results are interpreted and summarized using the big model to form a conclusion report;

[0046] Return the conclusion report to the user.

[0047] Optionally, the metadata management unit is further used to process the acquired insurance domain knowledge in blocks;

[0048] Vectorize the divided knowledge;

[0049] Create an index for the vectorized knowledge.

[0050] Optionally, the first data processing unit includes a context combination module, which is used to determine whether the current user query request is associated with a previous query request, and adjust the content of the model input based on the context information.

[0051] Optionally, the second data processing unit includes an answer generation module, which is used to generate a final user answer based on the first processing result generated by the first data processing unit in combination with context information, and return it to the user.

[0052] In a third aspect, the present application provides a device comprising a memory and a processor, wherein the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes the insurance data analysis method introduced in any implementation of the aforementioned second aspect.

[0053] In a fourth aspect, the present application provides a computer-readable storage medium having codes stored therein. When the codes are executed, a device executing the codes implements the insurance data analysis method described in any implementation of the aforementioned second aspect.

[0054] The present application provides an insurance data analysis system. Through the natural language processing technology of the first data processing unit, the system can accurately understand the user's query intention and automatically generate corresponding data query instructions. This greatly simplifies user operations, allowing even non-professional users to easily query data, thereby improving user experience. At the same time, the system can quickly respond to and process user requests, quickly generate a first processing result, and further improve work efficiency. In addition, the system is also equipped with multiple data analysis units, including a first data analysis unit, a second data analysis unit, and a third data analysis unit. These units can perform in-depth and comprehensive analysis of insurance data according to data query instructions. Through parallel processing or selective processing, the system can make full use of the advantages of each analysis unit and quickly obtain accurate analysis results. This multi-unit collaborative working mode significantly enhances the system's data processing and analysis capabilities. Finally, the first processing result is processed again by the second data processing unit to further refine and optimize the data and ensure the quality and accuracy of the final output result. This step helps to eliminate potential data errors or inconsistencies, making the output results more reliable and valuable for reference.

[0055] In summary, the insurance data analysis system significantly improves user experience and efficiency, enhances data processing and analysis capabilities, optimizes data output quality, and adapts to diversified analysis needs by integrating natural language processing technology and multiple data analysis units. These beneficial effects jointly promote the progress and development of the insurance data analysis field. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in this embodiment or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0057] Figure 1 A flowchart of an insurance data analysis method provided in an embodiment of the present application;

[0058] Figure 2 A schematic diagram of the structure of an insurance data analysis system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0060] Figure 1 A flowchart of an insurance data analysis method provided in an embodiment of the present application. Figure 1 As shown, the insurance data analysis method provided in the embodiment of the present application may include:

[0061] S101. Obtain a user's data query request, understand the user's query intent based on natural language processing technology, and generate a data query instruction.

[0062] The insurance data analysis method in this application is implemented based on a pre-built insurance data analysis system. The insurance system includes three intelligent agents, namely metadata encyclopedia knowledge, data query and extraction, and data analysis and interpretation.

[0063] The Metadata Encyclopedia Knowledge Agent is an agent that specializes in processing knowledge and information related to metadata in the insurance field. It is responsible for storing, managing, and retrieving data, knowledge, metadata information, etc. related to the insurance business. The agent can answer users' questions about metadata in the insurance field and provide relevant knowledge interpretation and information query services. It may use natural language processing technology to understand users' questions and provide accurate answers through retrieval or reasoning.

[0064] The data query and extraction agent is an agent responsible for querying and extracting data from the database. It generates corresponding query statements based on the user's query request and retrieves the required data from the database. The agent can parse the user's query intent, extract keywords, and generate corresponding SQL query statements or other data query languages. It can also handle complex data query requirements, such as multi-table association queries, data aggregation, etc., and present the query results in a user-friendly way.

[0065] The data analysis and interpretation agent is an agent that analyzes and interprets insurance data. It uses data analysis algorithms and models to conduct in-depth analysis and mining of data extracted from the database. The agent is able to identify patterns and trends in the data and provide data-driven decision support. It is also able to generate easy-to-understand reports and visual charts to help users better understand the data and analysis results. In addition, the agent may also have the ability to predict and simulate data to predict future business trends and potential risks. These three agents together constitute the core part of the insurance data analysis system, and they work together to provide users with efficient and accurate insurance data analysis services.

[0066] When a user submits a query request to the insurance data analysis system, the system will first use its built-in large model to identify the intent and accurately determine the type of agent corresponding to the user's intent, such as an agent for metadata encyclopedia knowledge. In order to more accurately understand and handle the user's real problem, the system will perform necessary rewriting and expansion based on the query request entered by the user. If the system cannot clearly determine the user's intent, it will start the intent clarification link and confirm with the user through multiple rounds of dialogue to improve the accuracy of understanding. In addition, in the first data processing unit, the system is also equipped with a context combination module, which can intelligently determine whether the current query request is related to the previous request, and adjust the content of the model input accordingly to ensure the consistency and accuracy of the query. Once the user's intent is clear, the system will generate the corresponding data query instruction and send it to the corresponding agent for subsequent processing. Finally, the system will return the processing results of the agent to the user. These results may include data query results, data analysis reports, and visual charts, etc., aiming to provide users with comprehensive and accurate insurance data analysis services.

[0067] S102: Input the data query instruction into one or more of the first data analysis unit, the second data analysis unit and the third data analysis unit for processing to obtain a first processing result.

[0068] The first data analysis unit, the second data analysis unit and the third data analysis unit respectively include the metadata encyclopedia knowledge agent, the data query and extraction agent and the data analysis and interpretation agent, and the corresponding data analysis unit can realize the corresponding data analysis function through the agent.

[0069] Specifically, when a data query instruction is sent to the first data analysis unit, the system first vectorizes the instruction or directly searches based on the keywords in the instruction. Then, using the RAG (Retrieval Augmented Generation) method, the system searches and returns answers related to the user query from the metadata management unit.

[0070] If the data query instruction is sent to the second data analysis unit, the system will first extract the keywords in the instruction and generate the corresponding SQL query statement based on these keywords, or build a domain specific language (DSL) for query. Then, the generated query statement is executed in the database, and the query results are rendered as a visual component so that users can understand the data more intuitively.

[0071] When the data query instruction is sent to the third data analysis unit, the system will use the big model to extract the parameters and the instruction itself from the instruction. Then, the system will match the pre-prepared insurance data analysis general API and execute the corresponding API according to the extracted parameters and instructions. After obtaining the query data results, the big model will further interpret and summarize these results to form a conclusion report. Finally, the system returns the conclusion report to the user so that the user can make a decision based on the report.

[0072] Through such an operation process, the insurance data analysis system can efficiently handle the user's data query needs and call the corresponding data analysis unit and intelligent agent for processing according to the user's intention. At the same time, the system also provides functions such as visual display and conclusion report to help users better understand the data and analysis results.

[0073] S103: Process the data in the first processing result again to obtain an output result.

[0074] The first processing result obtained after one or more of the processing by the first data analysis unit, the second data analysis unit and the third data analysis unit needs to be processed again to obtain the final output result. Further processing of the data after the first processing result can ensure the accuracy and readability of the information while meeting the needs of users. The process includes rearranging the data and sorting or grouping the data according to the logical order of the data or user preferences. For example, events can be arranged in chronological order, or information can be sorted by importance. To make the data more orderly and easy to understand. It also includes data filtering to remove unnecessary or redundant information and improve the accuracy of the information. After the processing is completed, the processed data is embedded in a suitable context so that users can better understand the meaning and background of the data.

[0075] In one implementation of the embodiment of the present application, the second data processing unit is used to reprocess the data in the first processing result, and the second data processing unit includes an answer generation module. The function of this module is to generate a final user answer based on the first processing result and context information. The data in the first processing result is combined with the context information to form a complete information chain. Based on the integrated information, a clear answer to the user's question is generated.

[0076] In one implementation of the embodiment of the present application, the method further includes managing relevant data, knowledge, metadata information and analysis methods in the insurance field through a metadata management unit; and performing rapid retrieval and intelligent question and answering of data through the metadata management unit.

[0077] The metadata management unit is used to integrate and manage relevant data, knowledge, metadata information and analysis methods in the insurance field. This means that whether it is customer information, policy information, industry reports, or detailed terms of insurance products, risk assessment models, etc., they can all be uniformly included in the metadata management unit for orderly management and storage. This centralized management method not only improves the accessibility and security of data, but also facilitates subsequent data analysis and application.

[0078] In addition, the metadata management unit also has the ability of fast retrieval and intelligent question-answering. Through advanced retrieval algorithms and intelligent question-answering technology, users can quickly find the data and knowledge they need, and even get relevant answers and suggestions by asking questions in natural language. This function greatly improves the efficiency and accuracy of information retrieval, allowing users to make decisions and take actions faster.

[0079] Furthermore, in order to optimize data processing efficiency and retrieval performance, the metadata management unit also uses block processing and vectorized representation technology. Specifically, it first reasonably blocks the acquired insurance field knowledge and decomposes the huge knowledge system into small blocks that are easier to manage and process. Then, each block of knowledge is vectorized, that is, it is converted into a series of mathematical vectors so that the computer can process and calculate it more conveniently. Finally, an index is established for this vectorized knowledge so that relevant data and information can be quickly found when needed.

[0080] Taking the risk assessment of insurance products as an example, the metadata management unit can integrate and analyze a large amount of historical claims data, customer behavior data, and market trend information, and achieve rapid risk assessment and prediction through vectorized representation and index establishment. When insurance sales personnel need to assess the insurance risk of a potential customer, they only need to enter the relevant customer information in the metadata management unit, and the system can quickly give risk assessment results and suggestions, thereby greatly improving work efficiency and accuracy.

[0081] The above-mentioned embodiments of the present application significantly optimize the search and use process of metadata assets by creating three major intelligent entities, so that users can quickly obtain the required metadata in a convenient way of AI question-and-answer dialogue without relying on professional SQL skills. At the same time, it supports interactive data exploration and extraction, which greatly improves the user's data analysis efficiency. The intelligent analysis and interpretation function uses the big model to match the analysis method in the insurance field, calls the accurate API to obtain the analysis results, and then outputs the analysis summary through the big model, which further improves the user's analysis efficiency and accuracy. In addition, the big model technology is innovatively used to conduct question-and-answer data analysis in the insurance field, integrating the data asset search, understanding, exploration and analysis processes that originally needed to be completed in multiple modules into one tool, realizing the connection of workflows, thereby significantly improving the user experience and analysis efficiency. Compared with existing data analysis solutions, this application pays more attention to the combing and integration of insurance industry knowledge, as well as the refinement of special analysis perspectives and interpretation ideas for the insurance field. Through the intention understanding function of the big model, it can accurately identify the instructions corresponding to various analysis methods, directly call the DataAPI to output the analysis perspectives and conclusions exclusive to the business field, and then bring the results into the big model for summary. This approach not only meets business needs, but also ensures the accuracy and effectiveness of the analysis results, further improving the quality and value of data analysis.

[0082] The above are some specific implementations of an insurance data analysis method provided in the embodiment of the present application. Based on this, the present application also provides a corresponding device. The device provided in the embodiment of the present application will be introduced from the perspective of functional modularization.

[0083] Figure 2 A schematic diagram of the structure of an insurance data analysis system provided in an embodiment of the present application. Figure 2 As shown, the insurance data analysis system 200 provided in the embodiment of the present application includes:

[0084] A first data processing unit 210, a second data processing unit 220, a first data analysis unit 230, a second data analysis unit 240 and a third data analysis unit 250;

[0085] The first data processing unit 210 is used to receive a user's data query request, understand the user's query intention based on natural language processing technology, and generate a data query instruction; input the data query instruction into one or more of the corresponding first data analysis unit 230, second data analysis unit 240 and third data analysis unit 250 for processing to obtain a first processing result;

[0086] The second data processing unit 320 is used to re-process the data in the first processing result to obtain the final output result.

[0087] In one implementation of the embodiment of the present application, the system further includes:

[0088] The metadata management unit is used to store and manage relevant data, knowledge, metadata information and analysis methods in the insurance field, and supports rapid data retrieval and intelligent question and answer.

[0089] In an implementation manner of the embodiment of the present application, the first data analysis unit is specifically used to:

[0090] Vectorizing the data query instruction, and performing a search based on the vectorized data, or directly performing a keyword search based on the data query instruction;

[0091] The answer is queried from the metadata management unit by using the retrieval enhancement to generate RAG.

[0092] In an implementation manner of the embodiment of the present application, the second data analysis unit is specifically configured to:

[0093] Extracting keywords from the data query instruction;

[0094] Generate corresponding SQL query statements based on the extracted keywords or compose domain-specific language DSL for query;

[0095] Execute query operations in the database based on the generated query statements, and render the query results as visual components for display.

[0096] In an implementation of the embodiment of the present application, the third data analysis unit is specifically used to:

[0097] Extracting parameters and instructions from the data query instruction using a large model;

[0098] Match the prepared insurance data analysis general API and execute the relevant API according to the extracted parameters and instructions;

[0099] After obtaining the query data results, the query data results are interpreted and summarized using the big model to form a conclusion report;

[0100] Return the conclusion report to the user.

[0101] In an implementation of the embodiment of the present application, the metadata management unit further includes a knowledge segmentation module, a vectorization module and an index creation module, wherein:

[0102] The knowledge segmentation module is used to segment the acquired insurance domain knowledge;

[0103] The vectorization module is used to vectorize the knowledge after segmentation;

[0104] The index building module is used to build an index for the vectorized knowledge.

[0105] In one implementation of the embodiment of the present application, the first data processing unit includes a context combination module, which is used to determine whether the current user query request is associated with the previous query request, and adjust the content of the model input based on the context information.

[0106] In one implementation of an embodiment of the present application, the second data processing unit includes an answer generation module, which is used to generate a final user answer based on the first processing result generated by the first data processing unit in combination with context information, and return it to the user.

[0107] The embodiments of the present application also provide corresponding devices and computer storage media for implementing the solutions provided by the embodiments of the present application.

[0108] The device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes the method described in any embodiment of the present application.

[0109] The computer storage medium stores codes, and when the codes are executed, a device executing the codes implements the method described in any embodiment of the present application.

[0110] Through the description of the above implementation methods, it can be known that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a read-only memory (ROM) / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network communication device such as a router) to execute the methods described in each embodiment of the present application or some parts of the embodiments.

[0111] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0112] It should also be noted that the various embodiments in this specification are described in a progressive manner, and the same and similar parts between the various embodiments can refer to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and apparatus embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments. The device and apparatus embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components indicated as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0113] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An insurance data analysis system, characterized in that: The system comprises a first data processing unit, a second data processing unit, a first data analysis unit, a second data analysis unit and a third data analysis unit; The first data processing unit is used to receive a user's data query request, understand the user's query intention based on natural language processing technology, and generate a data query instruction; input the data query instruction into one or more of the corresponding first data analysis unit, second data analysis unit and third data analysis unit for processing to obtain a first processing result; The second data processing unit is used to reprocess the data in the first processing result to obtain the final output result.

2. The insurance data analysis system according to claim 1, characterized in that: The system further comprises: The metadata management unit is used to store and manage relevant data, knowledge, metadata information and analysis methods in the insurance field, and supports rapid data retrieval and intelligent question and answer.

3. The insurance data analysis system according to claim 2, characterized in that: The first data analysis unit is specifically used for: Vectorizing the data query instruction, and performing a search based on the vectorized data, or directly performing a keyword search based on the data query instruction; The answer is queried from the metadata management unit by using the retrieval enhancement to generate RAG.

4. The insurance data analysis system according to claim 1, characterized in that: The second data analysis unit is specifically used for: Extracting keywords from the data query instruction; Generate corresponding SQL query statements based on the extracted keywords or compose domain-specific language DSL for query; Execute query operations in the database based on the generated query statements, and render the query results as visual components for display.

5. The insurance data analysis system according to claim 1, characterized in that: The third data analysis unit is specifically used for: Extracting parameters and instructions from the data query instruction using a large model; Match the prepared insurance data analysis general API and execute the relevant API according to the extracted parameters and instructions; After obtaining the query data results, the query data results are interpreted and summarized using the big model to form a conclusion report; Return the conclusion report to the user.

6. The insurance data analysis system according to claim 2, characterized in that: The metadata management unit also includes a knowledge segmentation module, a vectorization module and an index creation module, wherein: The knowledge segmentation module is used to segment the acquired insurance domain knowledge; The vectorization module is used to vectorize the knowledge after segmentation; The index building module is used to build an index for the vectorized knowledge.

7. The insurance data analysis system according to claim 1, characterized in that: The first data processing unit includes a context combining module, which is used to determine whether the current user query request is associated with a previous query request and adjust the content of the model input based on the context information.

8. The insurance data analysis system according to claim 1, characterized in that: The second data processing unit includes an answer generation module, which is used to generate a final user answer based on the first processing result generated by the first data processing unit in combination with context information, and return it to the user.

9. An insurance data analysis method, characterized in that: The method comprises: Obtain the user's data query request, understand the user's query intent based on natural language processing technology, and generate data query instructions; Inputting the data query instruction into one or more of the first data analysis unit, the second data analysis unit and the third data analysis unit for processing to obtain a first processing result; The data in the first processing result is processed again to obtain the final output result.

10. A computing device, characterized in that The computing device includes: a memory and a processor; The memory is used to store computer programs; The processor is configured to implement the method according to claim 9 when executing the computer program.

Images