Knowledge base-based responsibility determination method, apparatus and device, and medium

Abstract

The invention relates to the technical field of artificial intelligence, can be applied to business scenes such as financial science and technology, medical health and the like, and discloses a knowledge base-based responsibility determination method, which comprises the following steps of: constructing a knowledge base comprising responsibility determination conditions, judgment basis and reference data, constructing an information retrieval module, extracting information required by responsibility determination through the module, and determining the responsibility of the knowledge base according to the extracted information. And generating core context data, and generating a responsibility affirmation result based on the reasoning template and the complete context data. According to the method, the responsibility affirmation result is automatically generated in an intelligent reasoning and information extraction mode, and complexity and subjectivity of traditional manual judgment are avoided; through efficient data retrieval and reasoning analysis, the accuracy and processing speed of the responsibility determination process are improved, the requirement of manual intervention is remarkably reduced, and the claim settlement efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a knowledge base-based responsibility identification method, device, equipment and storage medium. Background Art

[0002] In recent years, employer liability insurance has gradually gained attention around the world, especially in some industries with high risks and frequent accidents. As the government's supervision of corporate employer liability becomes increasingly strict, companies have chosen to purchase employer liability insurance to cope with the increasing labor safety risks. However, the existing employer liability insurance has many difficulties in the claims process, especially in the liability identification link.

[0003] First, the process of liability determination often requires a detailed investigation and analysis of the cause of the accident, which is not only time-consuming but also involves complex and cumbersome materials. Insurance companies need to conduct a comprehensive assessment of the materials provided by the injured and the employer to determine whether the accident meets the insurance terms and specific policy conditions. In this process, claims adjusters need to have extensive insurance product knowledge, industry experience, and professional judgment. However, in reality, many insurance companies have not established a standardized and automated liability determination process, resulting in low claims efficiency.

[0004] Secondly, the existing claims system mainly relies on manual operation. Especially when the number of cases increases or the turnover of claims personnel is high, manual judgment may have certain deviations or omissions, resulting in inaccurate or delayed claims results. The amount of information faced by claims personnel is huge and complex, and they often need to weigh and judge various factors, including whether the employer is at fault and whether the accident meets the compensation conditions, which greatly increases the difficulty of manual judgment.

[0005] In addition, as the types and situations of labor safety accidents continue to change, traditional manual judgment mechanisms are difficult to adapt to the diverse and complex claims needs. Especially in cross-industry and cross-regional claims cases, the differences in accident types and backgrounds make the claims process more challenging. The high mobility of claims personnel also means that experienced staff may leave or transfer to other positions at work, resulting in instability in the experience of claims judges, which to some extent affects the efficiency and accuracy of claims.

[0006] In the field of medical health, similar liability identification issues also exist, especially in the process of handling medical accidents and injury compensation, medical institutions and insurance companies face similar difficulties in liability identification. Factors such as medical staff's operational errors, equipment failures, and the patient's own health status often require complex investigations and diagnoses to determine whether there is negligence. The existing medical accident liability identification is also highly dependent on manual judgment and coordination of multi-party information, resulting in problems of insufficient processing efficiency and transparency.

[0007] In the financial sector, especially in property insurance and liability insurance, although there are certain automated claims systems, the determination of liability in complex cases still requires a lot of manual review and judgment. This manual intervention increases the risk of misjudgment, lengthens the case processing cycle, and reduces customer satisfaction.

[0008] In general, the existing employer liability insurance claims system has serious problems in the process of liability identification, such as inefficiency, manual misjudgment and insufficient information processing. Especially when the number of cases increases, the bottleneck of the claims process is particularly prominent. These problems need to be effectively solved. Summary of the invention

[0009] The main purpose of the present invention is to provide a knowledge base-based responsibility identification method, device, equipment and storage medium, aiming to solve the technical problem that the existing technology relies on manual judgment in the responsibility identification process, resulting in a complex, time-consuming and error-prone identification process, especially when faced with large amounts of data or complex scenarios, and efficiency and accuracy cannot be effectively guaranteed.

[0010] To achieve the above object, the present invention provides a knowledge base-based responsibility identification method, comprising:

[0011] Build a knowledge base containing responsibility identification conditions, judgment basis and reference data;

[0012] Constructing an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract the responsibility identification conditions and judgment basis in the knowledge base according to the input information of the responsibility identification task;

[0013] Based on the responsibility identification conditions and judgment basis in the knowledge base, construct a reasoning template for guiding the information retrieval module to perform reasoning analysis;

[0014] Obtaining input data related to the target responsibility identification task, extracting key information from the input data and transmitting it to the information retrieval module;

[0015] The information retrieval module generates core context data of the target responsibility identification task based on the key information, and retrieves supplementary information related to the target responsibility identification task from the knowledge base;

[0016] Associating the core context data with the supplementary information to generate complete context data;

[0017] The information retrieval module generates a responsibility determination result based on the reasoning template and complete context data.

[0018] Furthermore, to achieve the above-mentioned purpose, the present invention provides a knowledge base-based responsibility identification device, comprising:

[0019] The knowledge base construction module is used to construct a knowledge base containing responsibility identification conditions, judgment basis and reference data;

[0020] An information retrieval module, used to construct an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract responsibility identification conditions and judgment basis in the knowledge base according to input information of the responsibility identification task;

[0021] A reasoning template construction module, used to construct a reasoning template for guiding the information retrieval module to perform reasoning analysis based on the responsibility identification conditions and judgment basis in the knowledge base;

[0022] An input data processing module is used to obtain input data related to the target responsibility identification task, extract key information from the input data and transmit it to the information retrieval module;

[0023] A context data generation module, used for the information retrieval module to generate core context data of the target responsibility identification task according to the key information, and retrieve supplementary information related to the target responsibility identification task from the knowledge base;

[0024] A data association module, used to associate the core context data with the supplementary information to generate complete context data;

[0025] A responsibility determination result generation module is used for the information retrieval module to generate a responsibility determination result based on the reasoning template and complete context data.

[0026] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer device, which includes a memory, a processor, and a knowledge base-based responsibility identification program stored in the memory and executable on the processor, wherein the knowledge base-based responsibility identification program, when executed by the processor, implements the steps of the knowledge base-based responsibility identification method as described above.

[0027] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a knowledge base-based responsibility identification program is stored. When the knowledge base-based responsibility identification program is executed by a processor, the steps of the knowledge base-based responsibility identification method as described above are implemented.

[0028] Beneficial effects: The present invention relates to the field of artificial intelligence technology and can be applied to business scenarios such as financial technology and medical health. It discloses a method for liability identification based on a knowledge base, including: constructing a knowledge base containing liability identification conditions, judgment basis and reference data, constructing an information retrieval module and extracting information required for liability identification through the module, generating core context data, and generating liability identification results based on reasoning templates and complete context data. The present invention automatically generates liability identification results through intelligent reasoning and information extraction, avoiding the complexity and subjectivity of traditional manual judgment; through efficient data retrieval and reasoning analysis, it improves the accuracy and processing speed of the liability identification process, significantly reduces the need for manual intervention, and improves the efficiency and accuracy of claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0030] Figure 1 A schematic diagram of an application environment of a knowledge base-based responsibility identification method in an embodiment of the present invention;

[0031] Figure 2 It is a flow chart of an embodiment of a knowledge base-based responsibility identification method of the present invention;

[0032] Figure 3 It is a functional module diagram of a preferred embodiment of the knowledge base-based responsibility identification device of the present invention;

[0033] Figure 4 A schematic diagram of the structure of a computer device in one embodiment of the present invention;

[0034] Figure 5 FIG. 4 is another schematic diagram of the structure of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0035] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0036] The knowledge base-based responsibility identification method provided by the embodiment of the present invention can be applied in the following aspects: Figure 1In an application environment, the user terminal communicates with the server terminal through a network. The server terminal can construct a knowledge base including liability identification conditions, judgment basis and reference data through the user terminal, construct an information retrieval module and extract the information required for liability identification through the module, generate core context data, and generate liability identification results based on the reasoning template and the complete context data. The present invention automatically generates liability identification results through intelligent reasoning and information extraction, avoiding the complexity and subjectivity of traditional manual judgment; through efficient data retrieval and reasoning analysis, the accuracy and processing speed of the liability identification process are improved, the need for manual intervention is significantly reduced, and the efficiency and accuracy of claims are improved. Among them, the user terminal can be but is not limited to various personal computers, laptops, smart phones, tablet computers and portable wearable devices. The server can be implemented with an independent server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0037] See also Figure 2 , Figure 2 This is a flow chart of an embodiment of a knowledge base-based responsibility identification method provided by the present invention. It should be noted that although a logical sequence is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0038] like Figure 2 As shown, the knowledge base-based responsibility identification method proposed by the present invention includes the following steps:

[0039] S10, building a knowledge base containing responsibility identification conditions, judgment basis and reference data;

[0040] In this embodiment, a knowledge base is constructed, which includes various conditions, judgment bases and related reference data required in the process of responsibility identification. Responsibility identification conditions usually refer to various factors that need to be considered when determining whether a certain responsibility is established, such as legal terms, contractual provisions, etc.; judgment bases are the rules or standards based on which responsibility identification is carried out according to these conditions; reference data include historical data, case information, etc., which help to provide support for responsibility identification.

[0041] In the specific implementation, firstly, we collect relevant responsibility identification conditions and judgment basis through laws and regulations, industry standards, historical cases and other information. In order to ensure the integrity and accuracy of the knowledge base, we can classify the knowledge base data according to different responsibility identification scenarios to form responsibility identification templates in different fields and situations. These templates are convenient for subsequent reasoning analysis and information retrieval modules.

[0042] Extract and summarize specific data on liability determination from past historical cases. By analyzing these historical cases, extract the key conditions and basis for determining whether liability is established in similar cases, and archive them in the knowledge base. This process helps to build a preliminary knowledge base data set, so that the subsequent liability determination process can draw on these historical data to improve accuracy and efficiency.

[0043] By sorting out the data of closed liability identification cases, we extract relevant liability identification conditions, judgment basis and case background data, and use text mining and data analysis technology to structure this information and store it in the knowledge base. The key to this is to ensure that the extracted historical data can reflect a variety of complex situations and provide highly adaptable support for future liability identification.

[0044] By subdividing the conditions and basis for determining responsibility, different types of cases can be quickly matched to the most relevant conditions and basis for determination in specific scenarios, thereby improving the efficiency and accuracy of determining responsibility.

[0045] When building a knowledge base, the data in the knowledge base can be classified according to various factors such as industry, case type, regional laws, etc. For example, in the field of medical health, cases can be divided into categories such as medical accidents and drug liability, and the corresponding liability identification conditions and judgment basis are stored under each category. This modular storage method enables the information retrieval module to quickly locate the relevant knowledge base module in actual applications, reducing the interference of irrelevant information during the retrieval process.

[0046] Responsibility identification conditions, judgment basis and reference data are often interrelated. When dealing with responsibility identification tasks, relevant data must be able to quickly associate and work together. To this end, a reasonable data association strategy needs to be formulated when building a knowledge base.

[0047] By building a data relationship model, different categories of data are associated in the knowledge base. For example, the conditions for liability identification may involve a certain regulation, and the applicable conditions of the regulation are associated with specific judgment bases; similarly, reference data can be further refined into specific cases to help establish more accurate liability judgment standards. This association strategy can be implemented through the structured storage method of graph databases or relational databases.

[0048] After building the knowledge base, it is necessary to design an efficient retrieval mechanism so that the information retrieval module can quickly find relevant data from the knowledge base based on the input information. The design of a multi-dimensional retrieval mechanism allows retrieval from different angles and levels, which helps to improve the flexibility and accuracy of retrieval.

[0049] Design a multi-dimensional retrieval system based on content retrieval, keyword retrieval, and semantic retrieval. When implementing, you can combine machine learning technology so that the knowledge base can continuously optimize the retrieval strategy. For example, for different types of cases, the system can query based on the case's specific keywords, background information, legal provisions, and other dimensions, and return the relevant liability identification conditions and basis.

[0050] Example: In the financial field, building a knowledge base on liability identification for financial fraud, debt default and financial compliance can also provide financial institutions with accurate decision-making basis. For example, when a bank handles a loan default case, the claims adjuster can determine the attribution of loan default liability by searching the knowledge base for liability identification conditions, such as loan contract terms, default conditions, historical cases, etc. The knowledge base can store standards for calculating liquidated damages, borrower behavior norms and judicial precedents, helping staff to quickly determine whether the borrower has violated the contract terms and whether they need to bear compensation liability.

[0051] At the same time, the handling of financial fraud cases can also rely on the standardized processes and case libraries in the knowledge base. For suspected financial fraud cases, claims adjusters can determine how to handle the case by searching the knowledge base for information such as legal clauses, case background, and evidence requirements related to financial fraud. The knowledge base in the financial field integrates multiple scenarios and multi-dimensional retrieval mechanisms to help financial institutions make more accurate and timely liability determinations when facing high-risk loan or insurance cases, thereby effectively reducing risks, enhancing customer trust, and reducing legal disputes.

[0052] In the field of healthcare, building a knowledge base for medical accidents can provide medical insurance companies with more accurate basis for liability determination. When an insurance claim occurs, claims adjusters can quickly obtain relevant data by searching the knowledge base for the identification conditions, legal provisions and related cases of medical accidents to make more informed judgments. For example, for a claim case caused by a patient's surgical error, the knowledge base can provide information such as standard operating procedures, basis for negligence determination, historical cases, etc. related to the surgical accident. This knowledge base based on historical data and scenario-based modularization can help claims adjusters find the most relevant judgment criteria from massive information more quickly, thereby improving the accuracy and efficiency of claims. This not only improves the efficiency of claims, but also reduces the risk of disputes that may be caused by inaccurate judgments.

[0053] By building a multi-dimensional, classified and refined knowledge base, the accuracy and processing efficiency of responsibility identification tasks can be significantly improved. Combining historical cases with scenario-based modular data storage can help the executors of responsibility identification tasks quickly and accurately obtain the required identification conditions and judgment basis, thereby speeding up the claims decision-making process and reducing the interference of human subjective factors.

[0054] S20, constructing an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract responsibility identification conditions and judgment basis in the knowledge base according to input information of the responsibility identification task;

[0055] In this embodiment, the core function of the information retrieval module is to design a query engine to interact with the knowledge base, support the search of various needs or task information input by users, and extract relevant data or rules. This module not only performs basic queries, but also can handle more complex retrieval logic, such as fuzzy matching, conditional query and semantic analysis. The design of the information retrieval module should ensure that it has the ability to effectively retrieve different types of input information and can quickly and accurately obtain the required responsibility identification conditions and judgment basis from the knowledge base.

[0056] The query engine first receives the task information from the user, parses the information, and identifies the key information. By building a multi-dimensional retrieval interface, different types of queries can be supported, such as exact matching based on keywords or queries based on contextual semantics. When the user enters the task information, the module executes the query through the database access interface, initiates a request to the knowledge base, and obtains the corresponding rules, standards, and judgment basis. In addition, the information retrieval module also needs to have the ability to filter and sort the query results to ensure that the most relevant information is returned.

[0057] The information retrieval module needs to be able to extract relevant responsibility identification rules or standards based on the input information provided by the responsibility identification task. The input information may include the type of accident, the people involved, the time and location of the accident, etc. By analyzing this input information, the module can match the most suitable responsibility identification conditions and judgment basis from the knowledge base.

[0058] After receiving the input information of the task, the information retrieval module analyzes it and identifies the characteristics of key information in the task, such as accident type, relevant responsible persons, timestamp, etc. These characteristics will constitute the query conditions and perform retrieval in the knowledge base. Through rule matching or semantic analysis, the responsibility identification rules and judgment bases that match these conditions are extracted. This process not only relies on simple text matching, but may also involve complex semantic understanding technology to ensure that the task information is highly associated with the content in the knowledge base.

[0059] Example: In the financial field, the role of the information retrieval module is to help claims adjusters quickly extract liability identification conditions and judgment basis related to specific loan contracts from a large number of cases when faced with problems such as loan defaults. For example, in a case where a borrower defaults, the claims adjuster only needs to enter the borrower's credit information, default amount, and contract terms, and the information retrieval module can obtain relevant default liability determination rules and best practices in the financial industry from the knowledge base, providing an accurate basis for the final decision.

[0060] In the field of medical health, the information retrieval module can effectively support the determination of liability in medical accidents. For example, in a case where a patient was injured due to an operational error by a medical staff, the claims adjuster can input the patient's medical history, the doctor's operation records, and the specific circumstances of the accident. The information retrieval module will extract the medical liability determination conditions, medical negligence standards, and related judgment basis related to these input information from the knowledge base. This can help quickly assess whether the accident meets the claims conditions and improve the efficiency of claims.

[0061] By building an intelligent information retrieval module for accessing the knowledge base, the efficiency and accuracy of liability identification task processing can be greatly improved. This module not only reduces manual intervention and the probability of errors by accurately parsing task information and matching relevant liability identification conditions, but also greatly shortens the task processing time and improves the automation level of the claims process. Whether it is a simple query or a complex judgment, the information retrieval module can quickly provide the most relevant legal basis and standards, making liability identification more efficient and accurate.

[0062] S30, constructing a reasoning template for guiding the information retrieval module to perform reasoning analysis based on the responsibility identification conditions and judgment basis in the knowledge base;

[0063] In this embodiment, the reasoning template is a set of regularized and systematic logical frameworks designed based on the specific contents of the responsibility identification conditions and judgment basis, which aims to guide the information retrieval module on how to perform reasoning analysis based on the input task information, context data and existing knowledge base rules. The reasoning template can be regarded as an operation guide, which specifies the analysis logic and reasoning path of the data, ensuring that the information retrieval module can perform logical analysis steps in complex responsibility identification tasks, thereby providing more accurate results. Through the reasoning template, the system can flexibly apply the conditions and basis in the knowledge base to perform automated judgments, thereby improving the accuracy and efficiency of responsibility identification.

[0064] When constructing a reasoning template, we must first analyze the core features and logical requirements of the responsibility identification task. Through a deep understanding of the responsibility identification conditions and judgment basis in the knowledge base, we can design a template structure that includes key variables, logical relationships, and reasoning paths. The design of the reasoning template should take into account the diversity of different types of tasks, such as the standards for responsibility identification, changes in accident types, differences in judgment basis, etc. The template usually includes several sub-templates, which can be adjusted and reused for different responsibility identification scenarios. The reasoning template can also automatically generate the corresponding analysis steps and judgment criteria based on the specific context of the task, and assist the information retrieval module to accurately extract and analyze relevant data.

[0065] In some common responsibility identification tasks, the reasoning template can be designed in a standardized form. For example, in the responsibility identification task of an injury accident, the template can include the following key steps: analysis of the time and location of the accident, analysis of the identity and behavior of the relevant responsible persons, classification analysis of the cause of the accident, and judgment on whether it complies with the insurance terms. The template clearly specifies the input data and judgment criteria required for each step, and the information retrieval module can quickly perform reasoning analysis based on these criteria.

[0066] For complex or changeable responsibility identification tasks, the reasoning template can be designed to be adaptive. That is, the template dynamically adjusts the reasoning path according to different input data and specific situations. For example, for different types of occupational injury accidents, the template can select different judgment bases and reasoning paths according to the specific circumstances of the accident, ensuring that each task can be analyzed according to its unique context. This flexible reasoning template ensures that the system can still maintain efficient and accurate reasoning capabilities when dealing with highly complex cases.

[0067] The reasoning template is continuously optimized in a data-driven way. After each responsibility identification task is processed, the system can evaluate the effect of the reasoning template, feedback the analysis results, and then adjust and optimize the template. For example, some tasks may frequently have errors in the judgment process. By analyzing these errors, the system can intelligently adjust the weights or paths of certain conditions in the reasoning template, thereby gradually improving the reasoning accuracy of the template.

[0068] Example description: In the field of medical health, the application of reasoning templates can help quickly determine whether medical staff have committed medical negligence. For example, in a case where a patient's condition worsened due to a doctor's negligent diagnosis and treatment, the reasoning template will guide the information retrieval module to extract relevant information from the patient's medical records, the doctor's operation records and other data to analyze whether the occurrence of the accident meets the standards for medical liability identification. The template stipulates specific judgment steps, including the diagnosis of patient symptoms, the doctor's treatment records, the use of drugs, etc., and finally forms a clear conclusion on liability identification.

[0069] In the financial field, reasoning templates can be applied to the determination of liability in loan default cases. For example, in the case of a borrower defaulting, the reasoning template will analyze the terms of the loan contract, the specific circumstances of the default, and the borrower's payment history to determine whether the conditions for determining liability for default are met. The template will automatically select appropriate judgment basis based on different contract terms, and perform reasoning on this basis, and ultimately generate a determination result for loan default liability, helping financial institutions to handle the determination of default liability more quickly.

[0070] By constructing a reasoning template to guide the information retrieval module to perform reasoning analysis, the system's automated analysis capabilities can be effectively improved. The reasoning template provides a clear analysis path and rules for the information retrieval module, which can automatically perform systematic reasoning analysis in complex responsibility identification tasks and adjust the reasoning process according to the characteristics of the task. This not only reduces manual intervention and improves task processing efficiency, but also ensures the accuracy of judgment, ultimately achieving efficient and accurate responsibility identification results.

[0071] S40, obtaining input data related to the target responsibility identification task, extracting key information from the input data and transmitting it to the information retrieval module;

[0072] In this embodiment, obtaining input data related to the target responsibility identification task means that the system collects various information related to the responsibility identification task from multiple data sources, such as accident reports, medical records, financial statements, contract documents, etc. These input materials contain various types of information required for task execution, and the information may be structured, semi-structured or unstructured. Extracting key information refers to identifying data points or features directly related to the responsibility identification task from these input materials. Usually, this key information includes the time, place, information of the parties involved, accident description, cause of the accident, etc. The purpose of extracting key information is to remove redundant information and focus on core data that has a direct impact on responsibility identification. The extracted key information will be passed as input to the information retrieval module to provide support for subsequent analysis and reasoning.

[0073] When obtaining input data, first determine the data sources related to the task. For example, if the responsibility determination task involves accident investigation, the relevant input data may include accident reports, on-site monitoring data, and medical records of victims. The system automatically or manually collects these data from different sources to ensure the integrity and validity of the data.

[0074] The process of extracting key information can be achieved through a variety of technical means, such as text mining, natural language processing (NLP) technology, pattern recognition, etc. For structured data (such as tables in a database), extraction is usually based on predefined fields and conditions; for unstructured data (such as documents and pictures), the system uses text recognition technology (OCR) or semantic analysis technology to extract information that is critical to liability determination. The specific extraction process may include:

[0075] Extract basic information such as time, location, and responsible person from the accident report;

[0076] Extract key information such as injury type, injury severity, and treatment plan from medical records;

[0077] Extract liability clauses, agreed contents, etc. from contract documents.

[0078] The extracted key information will be delivered to the information retrieval module in a structured or standardized format. The delivery process includes packaging the extracted key information into a unified data format (such as JSON, XML, CSV, etc.) to ensure that the information retrieval module can correctly understand and process this data. The delivery of key information is the basis for subsequent information retrieval and reasoning analysis, ensuring that the information retrieval module can perform tasks in an accurate context.

[0079] For structured data, such as fields in an accident report form, the system can directly extract the required field information according to pre-set rules, such as the time, location, and parties involved in the accident. This information can be obtained through database queries or API interfaces, and then sorted and converted according to the rules and passed to the information retrieval module.

[0080] When processing unstructured data, such as contract documents or medical records, the system first converts images or scanned documents into actionable text data through optical character recognition (OCR) technology. Then, natural language processing (NLP) technology is used to perform semantic analysis on the text data to identify key content related to the responsibility identification task. For example, extracting liability-related clauses in a contract or extracting symptom descriptions in a medical record.

[0081] The target responsibility determination task may involve many different types of input data (such as medical records, contract terms, accident reports, etc.). The system needs to integrate data from different sources, unify the processing format and extract the corresponding key information. For example, if it is a medical accident responsibility determination, the system can simultaneously extract information such as medical diagnosis records and the time of the accident, and pass this key information to the information retrieval module.

[0082] Example: In the financial sector, the responsibility identification task may involve loan default cases. The system extracts key information such as loan amount, repayment period, default clause, etc. from the loan contract, and extracts relevant data such as repayment ability and capital flow from the borrower's financial statements. This key information will be passed to the information retrieval module to help the system analyze the responsibility for loan default and automatically determine whether the default responsibility should be borne by the borrower, guarantor or other relevant parties.

[0083] In the healthcare sector, liability determination tasks often involve the investigation of medical accidents. The system extracts key information from medical records, such as the patient's condition description, treatment records, surgical procedures, and medical staff's operation logs, to determine the cause of the accident and whether there is medical negligence. For example, the system may extract the patient's admission date, diagnosis results, surgical records, and postoperative recovery status from a medical record, and pass this data to the information retrieval module for subsequent analysis.

[0084] By efficiently acquiring and extracting key information from input materials related to the responsibility identification task, the need for manual intervention can be greatly reduced, and the efficiency and accuracy of responsibility identification can be improved. The extraction of key information provides accurate data support for subsequent information retrieval and reasoning analysis, ensuring that the system can quickly and accurately perform responsibility identification tasks. At the same time, the standardization and structured processing of information enables seamless connection of data from different sources, reducing the problems caused by data inconsistency.

[0085] S50, the information retrieval module generates core context data of the target responsibility identification task according to the key information, and retrieves supplementary information related to the target responsibility identification task from the knowledge base;

[0086] In this embodiment, the information retrieval module first determines the core context data of the target responsibility identification task based on the input key information (such as the time, location, responsible person, etc. in the accident event). Core context data refers to the background information that is most directly related to the target task. It is the basis for responsibility identification analysis, including but not limited to various facts, people, legal clauses, etc. directly related to the incident. For example, in accident claims, core context data may include the specific circumstances of the accident, the division of responsibilities between the parties to the accident, and relevant legal clauses. In the process of generating core context data, the information retrieval module will match, integrate and format relevant data based on the input key information to ensure that all important information is included.

[0087] After receiving the key information, the information retrieval module will first extract relevant information from the data source. For example, the system will find the corresponding accident report, medical records and related contract documents through the accident time and location mentioned in the key information, and then match them. Through matching, the system can identify core data points, such as the specific circumstances of the accident, the division of responsibilities between the victim and the perpetrator, etc.

[0088] When generating core context data, the information retrieval module organizes the collected relevant information into a certain format. The goal of formatting is to ensure that the data can be effectively passed to the subsequent analysis module to avoid data redundancy or inconsistency. For example, the system may organize information such as the time, location, responsible person, and legal terms of the accident into a structured data block for subsequent use.

[0089] After generating the core context data, the information retrieval module needs to retrieve supplementary information related to the responsibility determination task from the knowledge base. This supplementary information may include previous similar cases, relevant laws and regulations, insurance clauses, industry standards, etc. The role of this supplementary information is to help enhance the understanding of the core context data and provide more comprehensive support for responsibility determination.

[0090] The information retrieval module initiates a retrieval request to the knowledge base based on the key information in the core context data. The retrieval content includes but is not limited to: the judgment results in historical cases, legal clauses or industry standards related to the target responsibility identification task, relevant provisions in insurance clauses, etc. Through retrieval, the system can obtain all supplementary information related to the target task, which provides more background basis for subsequent reasoning and analysis.

[0091] The supplementary information obtained from the knowledge base needs to be matched and integrated with the content in the core context data. Through matching, the system can filter out the most relevant supplementary information and exclude irrelevant content, thereby ensuring that subsequent analysis is based on the most valuable information. For example, when dealing with medical liability determination, the retrieved relevant legal provisions, industry standards, and judgments of similar cases will be combined with the core context data to help make a more accurate determination of liability.

[0092] When generating core context data, the system combines structured data (such as table information in a database) and unstructured data (such as text files and documents). For example, in an accident report, information about time, location, and responsible person is usually structured, while accident description and cause analysis may be unstructured. The information retrieval module needs to be able to process these two types of data and integrate them into core context data.

[0093] When retrieving supplementary information, the information retrieval module will not only retrieve traditional textual data, but may also retrieve data in other modalities such as images, videos, or voice. For example, in the task of determining responsibility for medical accidents, surveillance videos of the accident scene may become part of the supplementary information, and these videos need to use image recognition technology to extract valuable content to support subsequent reasoning analysis.

[0094] When searching for supplementary information, the information retrieval module may combine natural language processing technology or artificial intelligence algorithms to automatically optimize the search path based on the key information input. This intelligent retrieval can help the system extract relevant information from large amounts of data more efficiently and reduce the need for human intervention.

[0095] Example: In the healthcare field, suppose a medical accident occurs in a hospital and the patient demands that the hospital be held responsible. The information retrieval module will first extract key information such as the time, location, and personnel involved in the accident based on the accident report to generate core context data. Next, the system will retrieve cases and legal provisions related to similar medical accidents from the knowledge base to ensure the adequacy and relevance of the supplementary information. Finally, the system will combine the core context data and supplementary information to provide a basis for whether the hospital should assume responsibility.

[0096] In the financial field, if there is a task of determining the responsibility for loan default, the information retrieval module will first generate core context data based on key information such as the loan contract and the borrower's credit record. Then, the system will retrieve supplementary information such as relevant legal terms, historical cases, and financial industry standards from the knowledge base. By combining this supplementary information, the system can provide a more accurate analysis of the attribution of default responsibility and help financial institutions make reasonable decisions.

[0097] By generating core context data and retrieving supplementary information, the information retrieval module can provide accurate background data support for the responsibility determination task, ensuring that the task is performed based on sufficient information. This process not only improves the efficiency of responsibility determination, but also enhances its accuracy. By integrating core context data and supplementary information, the system can analyze the case more comprehensively and make more accurate judgments, thereby reducing the risk of human error and bias in the responsibility determination process.

[0098] S60, associating the core context data with the supplementary information to generate complete context data;

[0099] In this embodiment, the information retrieval module associates the generated core context data with the supplementary information retrieved from the knowledge base. The association process involves combining the information in the two data sets to ensure that the individual data points can be correctly matched, supplemented, and form a comprehensive context. This process is to ensure that the responsibility determination task can be analyzed based on a complete information flow, avoiding the omission of any key factors or missing information.

[0100] The association process first requires data alignment of the core context data and supplementary information. Since the sources and structures of the two may be different, the system needs to align the key information in the two through a data mapping algorithm. For example, the accident time in the core context data is compared with the applicable time period of the legal terms in the supplementary information to ensure the logical consistency between the information. Through data fusion technology, the system integrates the core data and supplementary information into a unified and comprehensive context to ensure that the analysis of the task has more dimensional support.

[0101] The system will establish association rules between core context data and supplementary information based on business rules or logical reasoning models. For example, when dealing with medical liability determination, the patient medical record information mentioned in the core context data and the medical diagnosis criteria in the supplementary information need to be associated according to specific rules. Through these rules, the system can provide more guiding and complete data for liability determination.

[0102] Complete context data refers to a comprehensive data set that can be used for reasoning analysis after associating core context data and supplementary information. This data set contains all necessary information and provides comprehensive support for subsequent responsibility determination decisions. The generated complete context data should be structured, highly operational and clear so that the subsequent reasoning process can proceed smoothly.

[0103] After the association is completed, the system will integrate the core contextual data with the supplementary information. The integrated data usually includes information from multiple dimensions, such as event description, responsibility allocation, applicable legal terms, historical case comparison, etc. This information will be unified into a structured data model to ensure that all data points are used reasonably and support subsequent responsibility identification analysis.

[0104] When generating complete contextual data, the system usually analyzes the data from multiple dimensions. These dimensions include legal provisions, historical cases, industry standards, facts of the event itself, etc. Depending on the task, the system will dynamically generate contextual data containing data from each dimension so that the reasoning analysis module can fully consider all relevant factors.

[0105] The system can design different data association models according to the needs of the actual task. For example, in some tasks, the timeline of the event is the most important, and the system will prioritize associating the time information in the core context data with the relevant legal clauses or cases in the supplementary information. For other tasks, it may be necessary to comprehensively associate various aspects of the event (such as the responsible person, scene description, etc.) to generate more complete background data.

[0106] During the association process, a rule-based engine can be introduced to dynamically determine which supplementary information should be matched with the core context data. These rules can be customized according to the specific circumstances of the task to improve the accuracy of information association. For example, some information should be combined with the core data only under certain conditions, while other information can be ignored.

[0107] Since the core contextual data and supplementary information may come from different data sources, the system needs to use data cleaning and preprocessing techniques to ensure that there is no redundant or contradictory content between the two data sets. Only after cleaning can the data be effectively associated and generate complete contextual data.

[0108] Example: In the healthcare field, suppose a patient files a claim for compensation due to a hospital surgical error. The information retrieval module first generates core context data through the accident report, including the time, location, relevant medical personnel, accident description, etc. of the accident. Then, the module retrieves relevant legal provisions, industry standards, and judgments of similar cases from the knowledge base. By associating this supplementary information with the core context data, the system is able to generate a comprehensive context data to provide a basis for whether the hospital should bear responsibility.

[0109] In the financial field, suppose a borrower fails to repay on time, resulting in loan default. The information retrieval module first generates core context data based on the loan contract and the borrower's credit information. Then, relevant financial legal terms, historical cases, and industry norms are retrieved from the knowledge base. By associating this supplementary information with the core context data, the system is able to generate a complete context data to help financial institutions determine the attribution of the borrower's default liability and make appropriate decisions.

[0110] By effectively associating core contextual data with supplementary information and generating complete contextual data, the information retrieval module can provide more comprehensive and accurate support. This process enables the responsibility determination task to be based not only on the current core information, but also on various supplementary information such as historical cases, legal provisions, and industry standards, thereby providing a more accurate and complete basis for judgment. This helps to improve the accuracy of responsibility determination and the rationality of decision-making, and reduce human bias and omissions.

[0111] S70, the information retrieval module generates a responsibility determination result based on the reasoning template and complete context data.

[0112] In this embodiment, the information retrieval module plays a core role in the responsibility identification process. Its main task is to extract relevant responsibility identification conditions, judgment basis and supplementary information from the knowledge base. In the process of generating the responsibility identification result, the information retrieval module will perform reasoning analysis based on the reasoning template and complete context data. This module can integrate a large language model and configure its analysis and matching strategy to ensure that it can accurately extract information related to responsibility identification from a large amount of knowledge base data.

[0113] Integrated large language model: The information retrieval module integrates a large language model (such as the Transformer architecture) for natural language understanding and information retrieval. The model can extract responsibility identification conditions and judgment basis through context analysis, and perform reasoning based on a given reasoning template.

[0114] Reasoning template application: The reasoning template sets rules and frameworks to guide the large language model on how to perform logical reasoning on input information. The reasoning template generates actionable identification results by performing structured analysis on the input information.

[0115] Contextual data generation and matching: The complete contextual data consists of information extracted from the knowledge base and key information in the input data. The information retrieval module will analyze this data, identify the key factors in responsibility determination, and finally derive the conclusion of responsibility determination.

[0116] Reasoning analysis is one of the important functions of the information retrieval module. It uses logical reasoning based on reasoning templates and complete contextual data to draw the conclusion of responsibility determination. Reasoning analysis is not just about matching and extracting data. It involves causal reasoning based on historical cases, rules and judgment basis to draw the final conclusion of responsibility determination.

[0117] Reasoning analysis process: The information retrieval module first analyzes the input data (such as the injured person's information, accident process, etc.) based on the rules defined in the reasoning template, and extracts the conditions related to the responsibility determination from the context data. Then, through the logical rules in the reasoning template, the relevant information is analyzed for association, and finally a reasonable responsibility determination result is generated.

[0118] Feedback mechanism: During the process of reasoning analysis, the information retrieval module will verify the reasoning results according to certain standards, identify potential anomalies, and optimize the reasoning results through the feedback mechanism.

[0119] Based on the complete contextual data and reasoning templates, the information retrieval module generates a liability determination result. This result is usually based on a series of logical reasoning and involves a comprehensive evaluation of multiple factors, including the cause of the accident, the division of liability, etc. The generated determination result will provide an important basis for claims decision-making.

[0120] Information extraction and processing: The information retrieval module extracts key information based on the results of reasoning and analysis, and processes this data through algorithms to form a concise and clear conclusion on responsibility determination.

[0121] Data output and result display: The generated liability determination results can be displayed in the form of a report, or directly sent to the claims system, interactive interface and other platforms for subsequent decision-making. The results should have a clear division of responsibilities and provide the basis for supporting decision-making.

[0122] Example description: In the financial field, the information retrieval module can be applied to the liability determination process in insurance claims. For example, for auto insurance claims, the information retrieval module can generate a liability determination result based on input data such as accident reports, car owner information, and eyewitness testimony, through reasoning and analysis, to determine whether the insurance compensation conditions are met. The system can help insurance companies efficiently and accurately determine accident liability through rapid reasoning and analysis, thereby making claims decisions.

[0123] In the field of medical health, the information retrieval module can be applied to the scene of medical liability identification. For example, when a medical accident occurs, the information retrieval module can analyze whether the accident is caused by medical operation errors based on the patient's medical records, doctor's operation records, drug use records and other input data through the reasoning template. Ultimately, the identification result of whether medical liability should be borne is generated, and support is provided for the subsequent compensation plan.

[0124] By integrating reasoning templates with contextual data, the information retrieval module can generate liability determination results more accurately. When dealing with complex liability determination tasks, this technology can automatically analyze large amounts of data and provide a basis for supporting decision-making, reduce the subjectivity of manual judgment, and improve work efficiency and accuracy. In addition, the automated liability determination process can significantly reduce the risk of claims caused by an increase in caseload or personnel turnover.

[0125] The present invention relates to the field of artificial intelligence technology, and can be applied to business scenarios such as financial technology and medical health. A knowledge base-based responsibility identification method is disclosed, including: constructing a knowledge base containing responsibility identification conditions, judgment basis and reference data, constructing an information retrieval module and extracting information required for responsibility identification through the module, generating core context data, and generating responsibility identification results based on reasoning templates and complete context data. The present invention automatically generates responsibility identification results through intelligent reasoning and information extraction, avoiding the complexity and subjectivity of traditional manual judgment; through efficient data retrieval and reasoning analysis, the accuracy and processing speed of the responsibility identification process are improved, the need for manual intervention is significantly reduced, and the efficiency and accuracy of claims are improved.

[0126] In one embodiment, the above S40 includes:

[0127] S401, when obtaining input data related to the target responsibility identification task, generating a responsibility identification task identifier based on metadata of the input data;

[0128] S402, extracting text information from the image in the input data by using a text recognition tool to generate text information;

[0129] S403, filtering key information for responsibility identification from the textual information, and storing the filtered key information in a preset data format;

[0130] S404, transferring the stored key information and the responsibility identification task identifier to the information retrieval module.

[0131] In this embodiment, the input data usually includes various types of data required for task execution (for example, accident reports, medical records, financial contracts, etc.), and the metadata describes the attributes of these input data, such as creation time, document source, file type, etc. By analyzing the metadata of the input data, a task identifier can be generated to uniquely identify a specific responsibility identification task. The task identifier will run through the entire responsibility identification process to ensure that all related data and operations are based on the same responsibility identification task.

[0132] The system generates a responsibility identification task ID by reading the metadata fields of the input data (such as file name, timestamp, data source, etc.) using a unique identifier (such as UUID). This ID is bound to the input data and stored in the task management system for subsequent retrieval and task tracking.

[0133] Input materials may include scanned documents, images or other unstructured data. In order to convert this data into actionable text information, the system uses text recognition tools (OCR) to process it. OCR technology can recognize text in images and convert it into text information that can be edited and analyzed.

[0134] The system integrates OCR tools (such as Tesseract, Baidu OCR API, etc.) to perform text recognition on images and extract the information in the image as text. During the extraction process, the recognition results will be preprocessed, such as removing noise and correcting text deviations to ensure the accuracy of the textual information.

[0135] Textual information may contain a large amount of irrelevant data. The system uses natural language processing (NLP) technology to filter out the core content related to liability identification, such as the time and location of the accident, the responsible party, and the loss description. The filtered key information needs to be stored in a predefined data format to facilitate subsequent retrieval and analysis.

[0136] The system extracts key information from textual information through methods such as keyword extraction and semantic analysis. The extracted content is stored in the database in a specified format (such as JSON, XML, CSV) to ensure the structure and uniformity of the data.

[0137] The filtered key information will be bound to the generated task identifier and used as input data for the information retrieval module, which relies on this data to perform subsequent analysis and reasoning.

[0138] The system encapsulates key information and task identifiers into standardized data packets, and uses data transmission protocols (such as HTTP API, message queues) to pass data to the information retrieval module to ensure the reliability and consistency of data transmission.

[0139] This embodiment effectively improves the efficiency and accuracy of the responsibility identification task by extracting key information from the input data and passing it to the information retrieval module. The task identifier is generated based on metadata to ensure the uniqueness and consistency of the task; key information is extracted through OCR and NLP technology to convert unstructured data into structured data, providing a reliable basis for subsequent analysis; at the same time, the standardized storage and transmission of information makes data processing more efficient and avoids the problem of information loss or inconsistent format.

[0140] In one embodiment, in the above S20, constructing an information retrieval module for accessing the knowledge base includes:

[0141] S201, integrating a large language model in the information retrieval module;

[0142] S202, configuring the analysis logic of the large language model so that the large language model can call the responsibility identification conditions and judgment basis in the knowledge base according to the input context data;

[0143] S203, setting a matching strategy of the large language model so as to be able to perform matching analysis between the input information and the data in the knowledge base when receiving the key information and the supplementary information;

[0144] S204, configuring the information retrieval module with a reasoning strategy for guiding the large language model to perform reasoning and association analysis on the input information and provide a matching result.

[0145] In this embodiment, the information retrieval module needs to integrate a large language model so that it can process, analyze and reason about natural language. The addition of a large language model enables the information retrieval module to not only perform information queries based on traditional keyword searches, but also perform complex matching and reasoning analysis based on semantic understanding. This feature greatly enhances the module's ability to handle complex tasks, such as accident descriptions in natural language or reasoning about liability identification conditions.

[0146] Deploy a large language model (such as a large model with a Transformer architecture) in the information retrieval module to support semantic parsing of input information. Configure an interactive interface with the large language model for the information retrieval module to ensure that the retrieval module can call on the model's reasoning capabilities and combine the output with the knowledge base content. Provide the model with adaptive computing resources (such as GPU acceleration) to ensure that it can run efficiently in large-scale tasks.

[0147] By setting up the analysis logic, the large language model can accurately extract key content when processing the input context data and automatically locate the conditions and basis related to liability determination in the knowledge base. This analysis logic enables the information retrieval module to conduct in-depth analysis of input information (such as accident reports or contract terms) and find the corresponding basis for judgment.

[0148] Build an analysis logic module and define the parsing rules for input information. For example, in the responsibility identification task, the analysis logic can be set to parse the accident time, responsible person, cause of the incident and other information. Configure the knowledge base call logic. Based on the keywords or semantic information in the context data, the system initiates a search request to the knowledge base to extract entries related to the responsibility identification conditions and judgment basis. Through the context understanding ability of the large language model, it is ensured that the input text information can be automatically converted into structured semantic data, so as to form an efficient match with the knowledge base content.

[0149] The role of the matching strategy is to guide the large language model on how to perform semantic matching and structural analysis on the input information and the data in the knowledge base. The matching strategy includes techniques such as keyword matching, semantic similarity calculation, and logical reasoning.

[0150] Configure keyword matching function: Design keyword extraction algorithm for large language model to ensure that the model can identify core keywords in input information and perform preliminary matching.

[0151] Integrated semantic similarity calculation: Utilize the semantic embedding characteristics of the model to calculate the semantic similarity between the input information and the knowledge base entries to ensure the accuracy of the matching results.

[0152] Design contextual matching rules: The matching strategy also needs to consider the context of the input information. The model further optimizes the matching results by analyzing the before and after logic of the input information.

[0153] The reasoning strategy provides a logical framework for the large language model, guiding it on how to associate input information with the knowledge base content and finally output matching results. This strategy includes not only simple data retrieval, but also condition-based logical reasoning and causal analysis.

[0154] Build an inference rule library: Define different inference rules according to the characteristics of different responsibility identification tasks (such as accident type, responsible party, etc.). The inference rule library guides the large language model on how to process input information.

[0155] Design reasoning paths: The reasoning strategy clearly specifies the selection logic of the reasoning path. For example, the system can select the corresponding liability identification conditions according to the accident type and perform step-by-step reasoning based on the context data.

[0156] Result output optimization: When generating matching results, the large language model optimizes the results according to the reasoning strategy to ensure that the output responsibility identification results are logical and actionable.

[0157] By integrating the large language model into the information retrieval module and configuring its analysis logic, matching strategy and reasoning strategy, this embodiment can significantly improve the automation of information processing and responsibility identification. The information retrieval module can not only quickly extract key data from complex contexts, but also provide accurate reasoning results according to task requirements. Compared with traditional manual analysis, this method can effectively reduce processing time, improve analysis accuracy, reduce error rate, and provide efficient support for responsibility identification tasks.

[0158] In one embodiment, the above S10 includes:

[0159] S101, collects responsibility identification conditions, judgment basis and reference data from historical responsibility identification cases to generate a preliminary knowledge base data set;

[0160] S102, according to different responsibility identification scenarios, the responsibility identification conditions and judgment bases in the preliminary knowledge base data set are divided and stored by category to construct a scenario-based knowledge base module;

[0161] S103, in the process of building the knowledge base, setting a data association strategy, and associating the responsibility identification conditions with relevant judgment bases and reference data according to the data association strategy;

[0162] S104, setting a multi-dimensional retrieval mechanism for the knowledge base that can perform retrieval according to responsibility scenarios, responsibility identification conditions or judgment bases.

[0163] In this embodiment, by collecting key content from historical liability identification cases, we extract liability identification conditions, judgment basis and related reference data as the basic data source of the knowledge base. The liability identification conditions in historical cases are usually clear rules or standards, the judgment basis may include legal provisions, industry norms, etc., and the reference data may include industry data, statistical data, etc.

[0164] The system automatically or manually extracts liability identification conditions, judgment basis and reference data from historical liability identification cases, such as collecting this information from judgment records and liability identification reports. Use natural language processing technology (NLP) to parse text data and convert historical case descriptions into structured conditions and basis. Store the extracted content as a preliminary knowledge base data set and create an index for subsequent processing for quick access.

[0165] Different liability identification scenarios (such as medical accidents, financial contract disputes, etc.) may require different liability identification conditions and judgment bases. By splitting storage by scenario, you can build a targeted scenario-based knowledge base module to ensure the manageability and availability of data. Classify the preliminary knowledge base data set according to the scenario classification rules. For example, data in the medical field is divided into diagnostic records, surgical records, etc., and data in the financial field is divided into contract terms, repayment records, etc. Use database technology to store data as different modules or tables. For example, one module stores medical-related rules and another module stores financial-related terms. Establish tags or metadata for the scenario-based knowledge base module so that it can quickly locate data in specific scenarios.

[0166] Data association strategy refers to the rules and methods for establishing links between data from different sources and types (such as responsibility identification conditions, judgment basis and reference data). Through this strategy, it can be ensured that the information in the knowledge base is linked to each other to form a complete identification logic chain.

[0167] Define data association rules, such as "the judgment basis corresponding to liability identification condition A includes legal clause B and industry specification C". Use relational database or graph database technology to store and index the relationships between different data nodes to ensure that related data can be quickly retrieved during query. The system can dynamically adjust the association strategy based on the logical relationship of the data, such as automatically updating the basis for association when updating legal clauses.

[0168] The multi-dimensional retrieval mechanism enables users to quickly query data in the knowledge base based on liability scenarios (such as medical and financial), liability identification conditions or judgment bases. This mechanism provides an efficient data access method for complex liability identification tasks.

[0169] Build a search engine that supports multi-dimensional queries, and supports searching by scenario, condition, or basis. Use full-text search technology (such as Elasticsearch) or multi-dimensional query functions based on databases to ensure that users can quickly locate the data they need. Design a friendly user interface to allow users to quickly complete queries by keyword, category label, or condition filtering.

[0170] This embodiment not only provides efficient data support for responsibility identification by constructing a knowledge base that includes responsibility identification conditions, judgment basis and reference data, but also significantly improves the efficiency and accuracy of task processing through scenario-based modules and multi-dimensional retrieval mechanisms. The scenario-based modular design of the knowledge base makes data easier to manage and use, the data association strategy ensures the logical integrity between conditions, basis and data, and the multi-dimensional retrieval mechanism allows users to quickly obtain relevant information to meet the needs of complex tasks.

[0171] In one embodiment, the above S50 includes:

[0172] S501, the information retrieval module generates core context data of the target responsibility identification task according to the key information;

[0173] S502, the information retrieval module determines a retrieval condition for supplementary information according to the core context data;

[0174] S503, the information retrieval module retrieves supplementary information related to the target responsibility identification task from the knowledge base according to the retrieval condition.

[0175] In this embodiment, core context data refers to the basic data set used for the responsibility identification task, including detailed parsing results of key information and background information related to the task. The information retrieval module generates core context data based on the input key information, combined with predefined logical rules and templates. These data can accurately reflect the specific background of the responsibility identification task and provide basic support for subsequent analysis.

[0176] After receiving key information (such as the time, location, responsible party, etc.), the system extracts specific information according to predefined parsing rules and structures it into core context data according to the template. Natural language processing (NLP) technology is used to further understand the semantic relationship of key information and integrate it into clear context data. Core context data may include timelines, parties involved, accident types, and related descriptions, which are stored in standardized formats (such as JSON or XML) for subsequent processing.

[0177] After generating the core context data, the information retrieval module needs to determine the retrieval conditions based on the content of the context data. The retrieval conditions are the bridge connecting the knowledge base content and the task requirements, and are used to define which information needs to be further obtained to support responsibility identification.

[0178] The system parses the core context data and extracts key fields (such as responsible parties, event descriptions, relevant time, etc.), which become the basic conditions for supplementary information retrieval. Use a dynamic rule generator to set the retrieval logic according to the characteristics of the task. For example, the system can define different retrieval scopes according to the type of accident (such as diagnosis and treatment specifications in the medical field and contract terms in the financial field). The system encapsulates the retrieval conditions into query statements and connects to the retrieval interface of the knowledge base to start the query process for supplementary information.

[0179] Supplementary information is an extension of the core contextual data, including historical cases, legal clauses, industry standards, etc. This information provides a more comprehensive background and basis for liability determination. The information retrieval module uses the retrieval interface of the knowledge base to query relevant data based on the set retrieval conditions, and screens and integrates the query results.

[0180] The system submits the search conditions to the knowledge base, triggering a multi-dimensional query. The query dimensions may include time, category, content keywords, etc. Use index optimization technology to improve search efficiency and ensure that supplementary information items matching the conditions are quickly returned. Filter and associate the search results to remove redundant data, ensure that the returned data is highly relevant, and form a logical association with the core context data.

[0181] In this embodiment, the information retrieval module can achieve comprehensive background data support for the task by generating core context data based on key information, defining supplementary information retrieval conditions, and retrieving supplementary information from the knowledge base. This ensures that the responsibility identification process is analyzed based on accurate and complete data, significantly improving the efficiency and accuracy of identification. The dynamically generated retrieval conditions improve the flexibility of the task, and the multi-dimensional retrieval mechanism ensures the comprehensiveness and relevance of the data, reducing information omissions and redundancy.

[0182] In one embodiment, the above S70 includes:

[0183] S701, the information retrieval module calls the reasoning template, analyzes the complete context data, and generates a preliminary responsibility determination result;

[0184] S702, verifying the preliminary responsibility determination result through the judgment standard set in the reasoning template, and identifying the existing abnormal elements in combination with the complete context data;

[0185] S703, generating prompt information associated with the responsibility determination result according to the abnormal factor;

[0186] S704, combining the prompt information with the preliminary responsibility determination result to generate a final responsibility determination result.

[0187] In this embodiment, in the preliminary stage of generating the responsibility identification result, the information retrieval module calls the reasoning template, analyzes the complete context data, extracts and processes key information, and generates a preliminary responsibility identification result according to the preset logic. This result may include a preliminary judgment and identification conclusion of the division of responsibilities as the basis for further verification.

[0188] The reasoning template contains logical rules and analysis frameworks, such as the criteria for determining liability conditions and the applicable rules for relevant evidence. The information retrieval module uses a large language model to parse complete context data (such as the cause of the accident, relevant legal provisions, etc.) and gradually derives preliminary identification results according to the logic of the reasoning template. The preliminary results are stored in a structured format (such as JSON or XML) to facilitate subsequent verification and prompt information generation.

[0189] The set of judgment criteria in the reasoning template is used to perform secondary verification on the preliminary identification results to ensure that the results are logical and consistent with the contextual data. The identification of abnormal elements is to find out potential problems that may affect the accuracy of responsibility identification, such as contradictory information, incomplete data, etc.

[0190] The system calls the verification rules in the reasoning template (for example, whether the negligence of the responsible party and the circumstances of the accident meet the clause standards) to verify the rationality of the preliminary results one by one. The identification of abnormal elements is achieved by comparing the logical consistency of the context data with the preliminary results. For example, it identifies whether there is missing information, judgments that contradict historical data, etc. The identified abnormal elements are recorded in the system log and passed to the next step to generate prompt information.

[0191] Prompt information is auxiliary content generated based on abnormal elements to help understand or correct the results of liability determination. These prompt information may include potential controversial points, uncertain factors or content that requires further verification.

[0192] The system generates prompt information based on the classification of abnormal elements, such as missing data prompts, inconsistent information prompts, or optional suggestions. Prompt information is generated based on preset templates and includes the type of abnormality, possible solutions, or suggestions for further analysis. Prompt information is attached to the responsibility identification results in the form of text descriptions for easy understanding and reference by users.

[0193] The final responsibility determination result is a complete output generated by combining the preliminary results and prompt information. This result includes a clear responsibility determination conclusion and related prompt information as the basis for task execution or subsequent processing.

[0194] The system integrates the prompt information into the relevant parts of the preliminary results and generates a responsibility determination result containing complete information. The final result is output in the form of a structured report (such as PDF or visual interface), which includes the determination conclusion, prompt information and supporting evidence. If the task requires, the final result can be passed to other system modules through the interface, such as risk control or decision support modules.

[0195] Example description: In the field of healthcare, the system receives a task to determine responsibility for a surgical error. The complete medical record data and diagnostic records are analyzed through reasoning templates to generate preliminary responsibility determination results, such as "improper operation during the operation." Subsequently, the system verifies the results according to the judgment criteria and finds that there is an abnormal element of inconsistent time in the surgical process recorded in the medical record. The prompt message reminds the user: "The time record of the surgical process does not match the description in other documents. Further verification is recommended." The final result combines the determination conclusion and prompt information to generate a complete responsibility determination report for reference by medical institutions or regulatory agencies.

[0196] In the financial sector, the system handles the task of determining the responsibility for a loan default. The system analyzes the loan contract and the customer's repayment record through the inference template to generate a preliminary result, such as "the customer's default responsibility is established." Subsequently, the system verifies the result and finds that the customer's repayment date record is inconsistent with the bank's archived information. The prompt message reminds the user: "The repayment date conflicts with the archived record. It is recommended to verify the archived information." The final responsibility determination result is generated in combination with the prompt information, including "the customer's default responsibility is established" and verification suggestions, providing detailed decision-making basis for financial institutions.

[0197] This embodiment significantly improves the automation level of responsibility identification and the accuracy of the results by calling the reasoning template, generating preliminary results and verifying them, and generating the final results in combination with prompt information. The verification mechanism effectively reduces logical errors or data omissions in the preliminary identification results, and the generation of prompt information provides users with more decision-making support. The final responsibility identification results are combined with prompt information to form a complete task report, which helps to improve the efficiency and reliability of subsequent operations.

[0198] In one embodiment, after the above S70, the method further includes:

[0199] S801, sending the responsibility determination result to the interactive interface;

[0200] S802, displaying the core context data, identification basis and the responsibility identification result of the target responsibility identification task in the interactive interface;

[0201] S803: configuring and displaying a review and feedback module in the interactive interface.

[0202] In this embodiment, after the responsibility determination result is generated, it needs to be sent to the interactive interface through the system's transmission module. The interactive interface is an important platform for users to view, review and provide feedback. The process of sending the responsibility determination result needs to ensure the integrity and timeliness of the data to support subsequent operations.

[0203] The system encapsulates the responsibility identification results into a standardized data package (such as JSON or XML format) and transmits it to the interactive interface module through the API interface or message queue. To ensure the integrity and security of data transmission, the system can use encrypted transmission (such as TLS protocol) and design a data verification mechanism. After the transmission is completed, the interactive interface module stores the responsibility identification results in the local cache or database and prepares for display.

[0204] The interactive interface must be able to clearly and structuredly display all key content related to the task, including core contextual data (such as accident background information), identification basis (such as judgment criteria and reference clauses), and responsibility identification results. This display method not only helps users quickly understand the identification logic, but also provides data support for review and feedback.

[0205] The interactive interface design can adopt a modular structure, presenting the core context data, identification basis and responsibility identification results in different display areas. The core context data can be displayed in card or table format, including key information such as the responsible party, accident time, and related events; the identification basis can directly reference the terms or standards in the knowledge base. The responsibility identification results must be presented with a clear conclusion (such as "responsibility established" or "responsibility not established"), and a detailed analysis report must be attached. To improve the user experience, the interactive interface can support dynamic display functions, allowing users to expand or hide specific information as needed.

[0206] The review and feedback module is an important function of the interactive interface. Users can review the responsibility determination results and provide feedback through this module. The review module helps users identify potential problems or provide additional information, while the feedback module records users' opinions and transmits them to the responsibility determination system.

[0207] The system provides a "Review" function button in the interactive interface. Users can click it to start the detailed analysis process of the responsibility identification task. It provides an interactive marking function, allowing users to mark the displayed identification basis or responsibility conclusion (such as marking "needs to be modified" or "needs further investigation"). The review module associates the user's marking record with the responsibility identification result and stores it in the system's log or task record.

[0208] The system sets a feedback input box in the interactive interface, allowing users to fill in opinions, suggestions or additional information. The feedback content is transmitted to the responsibility identification system through the API interface and stored together with the original task data for subsequent optimization or correction. The classification labels of the feedback data (such as "task information supplement" and "identification result questioning") are configured to facilitate the subsequent classification processing of the system.

[0209] Configure the permission control function to allow only authorized users to access the review and feedback module. Record user operations through logs to ensure the transparency and traceability of the responsibility identification system.

[0210] This embodiment significantly improves the efficiency of users understanding responsibility identification tasks by sending the responsibility identification results to the interactive interface and displaying relevant content. The configuration of the review and feedback module provides users with the ability to further verify and correct the responsibility identification results, enhancing the transparency and flexibility of task processing. It effectively reduces the time cost of the traditional manual verification process, ensures the accuracy of the responsibility identification results, and continuously optimizes the identification logic through user feedback to improve the intelligence level of the system.

[0211] In one embodiment, a knowledge base-based responsibility identification device is provided, and the knowledge base-based responsibility identification device corresponds one-to-one with the knowledge base-based responsibility identification method in the above embodiment. Figure 3 , Figure 3 This is a functional module diagram of a preferred embodiment of the knowledge base-based responsibility identification device of the present invention. The knowledge base construction module 10, the information retrieval module 20, the reasoning template construction module 30, the input data processing module 40, the context data generation module 50, the data association module 60 and the responsibility identification result generation module 70. The detailed description of each functional module is as follows:

[0212] The knowledge base construction module 10 is used to construct a knowledge base including responsibility identification conditions, judgment basis and reference data;

[0213] An information retrieval module 20, used to construct an information retrieval module for accessing the knowledge base, the information retrieval module is used to extract the responsibility identification conditions and judgment basis in the knowledge base according to the input information of the responsibility identification task;

[0214] A reasoning template construction module 30, for constructing a reasoning template for guiding the information retrieval module to perform reasoning analysis based on the responsibility identification conditions and judgment basis in the knowledge base;

[0215] An input data processing module 40 is used to obtain input data related to the target responsibility identification task, extract key information from the input data and transmit it to the information retrieval module;

[0216] A context data generation module 50, used for the information retrieval module to generate core context data of the target responsibility identification task according to the key information, and to retrieve supplementary information related to the target responsibility identification task from the knowledge base;

[0217] A data association module 60, configured to associate the core context data with the supplementary information to generate complete context data;

[0218] The responsibility determination result generating module 70 is used for the information retrieval module to generate the responsibility determination result based on the reasoning template and the complete context data.

[0219] In one embodiment, the input data processing module 40 is specifically used to:

[0220] When obtaining input data related to the target responsibility identification task, generating a responsibility identification task identifier based on metadata of the input data;

[0221] Extracting text information from the image in the input data by using a text recognition tool to generate text information;

[0222] Filtering key information for responsibility identification from the textual information, and storing the filtered key information in a preset data format;

[0223] The stored key information and the responsibility identification task identifier are transmitted to the information retrieval module.

[0224] In one embodiment, the information retrieval module 20 is specifically used to:

[0225] Integrate large language models in information retrieval modules;

[0226] Configuring the analysis logic of the large language model so that the large language model can call the responsibility identification conditions and judgment basis in the knowledge base according to the input context data;

[0227] Setting a matching strategy for the large language model so as to be able to perform matching analysis between input information and data in the knowledge base when receiving key information and supplementary information;

[0228] The information retrieval module is configured with a reasoning strategy for guiding the large language model to perform reasoning and association analysis on input information and provide matching results.

[0229] In one embodiment, the knowledge base construction module 10 is specifically used for:

[0230] Collect responsibility identification conditions, judgment basis and reference data from historical responsibility identification cases to generate a preliminary knowledge base data set;

[0231] According to different responsibility identification scenarios, the responsibility identification conditions and judgment bases in the preliminary knowledge base data set are split and stored by category to build a scenario-based knowledge base module;

[0232] In the process of building the knowledge base, a data association strategy is set, and the responsibility identification conditions are associated with relevant judgment bases and reference data according to the data association strategy;

[0233] A multi-dimensional retrieval mechanism is set up for the knowledge base, which can perform retrieval according to responsibility scenarios, responsibility identification conditions or judgment bases.

[0234] In one embodiment, the context data generating module 50 is specifically configured to:

[0235] The information retrieval module generates core context data of the target responsibility identification task based on the key information;

[0236] The information retrieval module determines the retrieval conditions for the supplementary information based on the core context data;

[0237] The information retrieval module retrieves supplementary information related to the target responsibility identification task from the knowledge base according to the retrieval condition.

[0238] In one embodiment, the responsibility determination result generating module 70 is specifically used to:

[0239] The information retrieval module calls the reasoning template to analyze the complete context data and generate a preliminary responsibility determination result;

[0240] Verify the preliminary responsibility determination result through the judgment standard set in the reasoning template, and identify the existing abnormal elements in combination with the complete context data;

[0241] Generating prompt information associated with the responsibility determination result according to the abnormal factors;

[0242] The prompt information is combined with the preliminary responsibility determination result to generate a final responsibility determination result.

[0243] In one embodiment, the responsibility determination result generating module 70 is specifically used to:

[0244] Sending the responsibility determination result to the interactive interface;

[0245] Displaying the core context data, identification basis and the responsibility identification result of the target responsibility identification task in the interactive interface;

[0246] A review and feedback module is configured and displayed in the interactive interface.

[0247] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external user terminal through a network connection. When the computer program is executed by the processor, it realizes the functions or steps of a knowledge-based responsibility identification method server side.

[0248] In one embodiment, a computer device is provided. The computer device may be a user terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server through a network connection. When the computer program is executed by the processor, it realizes the functions or steps of a user-side method of responsibility identification based on a knowledge base

[0249] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are implemented:

[0250] Build a knowledge base containing responsibility identification conditions, judgment basis and reference data;

[0251] Constructing an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract the responsibility identification conditions and judgment basis in the knowledge base according to the input information of the responsibility identification task;

[0252] Based on the responsibility identification conditions and judgment basis in the knowledge base, construct a reasoning template for guiding the information retrieval module to perform reasoning analysis;

[0253] Obtaining input data related to the target responsibility identification task, extracting key information from the input data and transmitting it to the information retrieval module;

[0254] The information retrieval module generates core context data of the target responsibility identification task based on the key information, and retrieves supplementary information related to the target responsibility identification task from the knowledge base;

[0255] Associating the core context data with the supplementary information to generate complete context data;

[0256] The information retrieval module generates a responsibility determination result based on the reasoning template and complete context data.

[0257] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0258] Build a knowledge base containing responsibility identification conditions, judgment basis and reference data;

[0259] Constructing an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract the responsibility identification conditions and judgment basis in the knowledge base according to the input information of the responsibility identification task;

[0260] Based on the responsibility identification conditions and judgment basis in the knowledge base, construct a reasoning template for guiding the information retrieval module to perform reasoning analysis;

[0261] Obtaining input data related to the target responsibility identification task, extracting key information from the input data and transmitting it to the information retrieval module;

[0262] The information retrieval module generates core context data of the target responsibility identification task based on the key information, and retrieves supplementary information related to the target responsibility identification task from the knowledge base;

[0263] Associating the core context data with the supplementary information to generate complete context data;

[0264] The information retrieval module generates a responsibility determination result based on the reasoning template and complete context data.

[0265] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can refer to the relevant descriptions on the server side and the user side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0266] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0267] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0268] It should be noted that if software tools or components other than those of the Company appear in the embodiments of the present application, they are only used for illustration and do not represent actual use. The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the above-mentioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the above-mentioned embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents; and these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A knowledge base-based responsibility identification method, characterized in that: The following steps are involved: Build a knowledge base containing responsibility identification conditions, judgment basis and reference data; Constructing an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract the responsibility identification conditions and judgment basis in the knowledge base according to the input information of the responsibility identification task; Based on the responsibility identification conditions and judgment basis in the knowledge base, construct a reasoning template for guiding the information retrieval module to perform reasoning analysis; Obtaining input data related to the target responsibility identification task, extracting key information from the input data and transmitting it to the information retrieval module; The information retrieval module generates core context data of the target responsibility identification task based on the key information, and retrieves supplementary information related to the target responsibility identification task from the knowledge base; Associating the core context data with the supplementary information to generate complete context data; The information retrieval module generates a responsibility determination result based on the reasoning template and complete context data.

2. The knowledge-based responsibility identification method according to claim 1, characterized in that: Obtaining input data related to the target responsibility identification task, extracting key information from the input data and transmitting it to the information retrieval module, including: When obtaining input data related to the target responsibility identification task, generating a responsibility identification task identifier based on metadata of the input data; Extracting text information from the image in the input data by using a text recognition tool to generate text information; Filtering key information for responsibility identification from the textual information, and storing the filtered key information in a preset data format; The stored key information and the responsibility identification task identifier are transmitted to the information retrieval module.

3. The knowledge-based responsibility identification method according to claim 1, characterized in that: Constructing an information retrieval module for accessing the knowledge base, including: Integrate large language models in information retrieval modules; Configuring the analysis logic of the large language model so that the large language model can call the responsibility identification conditions and judgment basis in the knowledge base according to the input context data; Setting a matching strategy for the large language model so as to be able to perform matching analysis between input information and data in the knowledge base when receiving key information and supplementary information; The information retrieval module is configured with a reasoning strategy for guiding the large language model to perform reasoning and association analysis on input information and provide matching results.

4. The knowledge-based responsibility identification method according to claim 1, characterized in that: Build a knowledge base containing responsibility identification conditions, judgment basis and reference data, including: Collect responsibility identification conditions, judgment basis and reference data from historical responsibility identification cases to generate a preliminary knowledge base data set; According to different responsibility identification scenarios, the responsibility identification conditions and judgment bases in the preliminary knowledge base data set are split and stored by category to build a scenario-based knowledge base module; In the process of building the knowledge base, a data association strategy is set, and the responsibility identification conditions are associated with relevant judgment bases and reference data according to the data association strategy; A multi-dimensional retrieval mechanism is set up for the knowledge base, which can perform retrieval according to responsibility scenarios, responsibility identification conditions or judgment bases.

5. The knowledge-based responsibility identification method according to claim 1, characterized in that: The information retrieval module generates core context data of the target responsibility identification task according to the key information, and retrieves supplementary information related to the target responsibility identification task from the knowledge base, including: The information retrieval module generates core context data of the target responsibility identification task based on the key information; The information retrieval module determines the retrieval conditions for the supplementary information based on the core context data; The information retrieval module retrieves supplementary information related to the target responsibility identification task from the knowledge base according to the retrieval condition.

6. The knowledge-based responsibility identification method according to claim 1, characterized in that: The information retrieval module generates a responsibility determination result based on the reasoning template and the complete context data, including: The information retrieval module calls the reasoning template to analyze the complete context data and generate a preliminary responsibility determination result; Verify the preliminary responsibility determination result through the judgment standard set in the reasoning template, and identify the existing abnormal elements in combination with the complete context data; Generating prompt information associated with the responsibility determination result according to the abnormal factors; The prompt information is combined with the preliminary responsibility determination result to generate a final responsibility determination result.

7. The knowledge-based responsibility identification method according to claim 1, characterized in that: After the information retrieval module generates the responsibility determination result based on the reasoning template and the complete context data, it also includes: Sending the responsibility determination result to the interactive interface; Displaying the core context data, identification basis and the responsibility identification result of the target responsibility identification task in the interactive interface; A review and feedback module is configured and displayed in the interactive interface.

8. A knowledge base-based responsibility identification device, characterized in that: The knowledge base-based responsibility identification device includes: The knowledge base construction module is used to construct a knowledge base containing responsibility identification conditions, judgment basis and reference data; An information retrieval module, used to construct an information retrieval module for accessing the knowledge base, wherein the information retrieval module is used to extract responsibility identification conditions and judgment basis in the knowledge base according to input information of the responsibility identification task; A reasoning template construction module, used to construct a reasoning template for guiding the information retrieval module to perform reasoning analysis based on the responsibility identification conditions and judgment basis in the knowledge base; An input data processing module is used to obtain input data related to the target responsibility identification task, extract key information from the input data and transmit it to the information retrieval module; A context data generation module, used for the information retrieval module to generate core context data of the target responsibility identification task according to the key information, and retrieve supplementary information related to the target responsibility identification task from the knowledge base; A data association module, used to associate the core context data with the supplementary information to generate complete context data; A responsibility determination result generation module is used for the information retrieval module to generate a responsibility determination result based on the reasoning template and complete context data.

9. A computer device, characterized in that: The computer device includes a memory, a processor, and a knowledge-base based responsibility identification program stored in the memory and executable on the processor. When the knowledge-base based responsibility identification program is executed by the processor, the steps of the knowledge-base based responsibility identification method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The storage medium stores a knowledge-based responsibility identification program, which, when executed by a processor, implements the steps of the knowledge-based responsibility identification method as described in any one of claims 1 to 7.

Images