Method and related device for claim analysis of car insurance claim settlement case

Abstract

The invention discloses a method for carrying out claim analysis on a car insurance claim settlement case and a related device, and relates to the technical field of car insurance claim settlement. Comprising the steps of obtaining business data of a target car insurance claim settlement case; and inputting the business data and the target prompt word into a case analysis model, analyzing the business data through the case analysis model, and directly outputting an analysis result of a target car insurance claim settlement case corresponding to the target prompt word. Therefore, the time spent on detailed analysis of each case by front-line service personnel in a traditional scheme is saved, the analysis result of the car insurance claim settlement case is directly output through the case analysis model, the work efficiency of claim analysis of the car insurance claim settlement case is greatly improved, and the work pressure of the front-line service personnel is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of automobile insurance claims, and in particular to a method and related device for performing compensation analysis on automobile insurance claims cases. Background Art

[0002] Auto insurance claims refer to a series of processes and results in which the insurer handles the claim filed by the insured in accordance with the insurance contract after an accident occurs within the scope of insurance liability. In traditional solutions, front-line business personnel need to spend a lot of time sorting out the facts of vehicle claims, combining insurance product conditions and traffic laws and regulations to determine which insurance clauses of the customer in the auto insurance claims case can be compensated, whether the deductible clause is triggered, and what the compensation ratio is.

[0003] With the improvement of living standards, the number of auto insurance claims cases has continued to increase, and front-line operators have long been overwhelmed. How to reduce the work pressure of front-line operators while ensuring efficient compensation analysis of auto insurance claims cases has become one of the technical issues that need to be urgently solved in the field of auto insurance claims technology. Summary of the invention

[0004] Based on the above problems, the present application provides a method for performing compensation analysis on motor insurance claim cases, which is used to perform efficient compensation analysis on motor insurance claim cases and reduce the work pressure of front-line operators.

[0005] The embodiments of the present application disclose the following technical solutions:

[0006] The first aspect of the present application discloses a method for performing compensation analysis on a motor vehicle insurance claim case, comprising:

[0007] Obtain business data of target auto insurance claims cases;

[0008] Through the case analysis model, according to the business data and the target prompt word, the analysis results of the target motor insurance claim case corresponding to the target prompt word are obtained; the case analysis model is a model pre-trained based on a large language model, and is used to output the analysis results of the motor insurance claim case corresponding to the prompt word.

[0009] In an optional implementation, the case analysis model includes an analysis module, an action module and an evaluation module; the case analysis model is used to obtain, according to the business data and the target prompt word, an analysis result of a target auto insurance claim case corresponding to the target prompt word, including:

[0010] Generate, by the analysis module, either a tool calling scheme or a preliminary analysis result according to the business data and the target prompt word;

[0011] If the preliminary analysis result is output, the preliminary analysis result is used as the analysis result;

[0012] If the tool calling scheme is output, the analysis result is generated according to the business data, the target prompt word and the tool calling scheme through the action module, the evaluation module and the analysis module.

[0013] In an optional implementation, the generating of the analysis result according to the business data, the target prompt word and the tool calling scheme by the action module, the evaluation module and the analysis module includes:

[0014] Through the action module, the tool in the tool calling scheme is called to obtain a calling result; the calling result includes: a summary of traffic laws, insurance clauses or the target auto insurance claim case;

[0015] By using the evaluation model, the call result is analyzed to obtain a judgment result; the judgment result includes whether it is necessary to obtain supplementary information for analyzing the target auto insurance claim case, and the name of the supplementary information;

[0016] Through the analysis module, an updated tool calling scheme or the analysis result is generated based on the judgment result, the business data and the target prompt word; the updated tool calling scheme is used as input data of the evaluation module to generate the analysis result.

[0017] In an optional implementation, the training step of the case analysis model includes:

[0018] Acquire multiple groups of sample data, each group of sample data includes: historical business data of historical motor insurance claims cases and analysis results of the historical motor insurance claims cases;

[0019] The large language model to be trained is trained based on the sample prompt words and the multiple groups of sample data to obtain the case analysis model.

[0020] In an optional implementation, the business data includes: survey conversation records, insurance data purchased by the user, insurance product terms or accident liability.

[0021] The second aspect of the present application discloses a device for performing compensation analysis on a motor vehicle insurance claim case, comprising:

[0022] Data acquisition module, used to obtain business data of target auto insurance claims cases;

[0023] The analysis result acquisition module is used to obtain the analysis results of the target motor vehicle insurance claim case corresponding to the target prompt word through the case analysis model according to the business data and the target prompt word; the case analysis model is a model pre-trained based on the large language model and is used to output the analysis results of the motor vehicle insurance claim case corresponding to the prompt word.

[0024] In an optional implementation, the analysis result acquisition module includes:

[0025] A data analysis unit, configured to generate, through the analysis module, any one of a tool calling scheme and a preliminary analysis result according to the business data and the target prompt word;

[0026] A first analysis result obtaining unit, configured to use the preliminary analysis result as the analysis result if the preliminary analysis result is output;

[0027] The second analysis result acquisition unit is used to generate the analysis result according to the business data, the target prompt word and the tool calling plan through the action module, the evaluation module and the analysis module if the tool calling plan is output.

[0028] In an optional implementation, the second analysis result obtaining unit includes:

[0029] A call result acquisition subunit is used to call the tool in the tool call scheme through the action module to obtain a call result; the call result includes: a summary of traffic laws, insurance clauses or the target auto insurance claim case;

[0030] A judgment result acquisition subunit is used to analyze the call result through the evaluation model to obtain a judgment result; the judgment result includes whether it is necessary to obtain supplementary information for analyzing the target motor insurance claim case, and the name of the supplementary information;

[0031] The second analysis result acquisition subunit is used to generate an updated tool calling scheme or the analysis result through the analysis module based on the judgment result, the business data and the target prompt word; the updated tool calling scheme is used as input data of the evaluation module to generate the analysis result.

[0032] A third aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method introduced in any implementation manner of the first aspect.

[0033] A fourth aspect of the present application provides an electronic device, including:

[0034] a memory having a computer program stored thereon;

[0035] A processor is used to execute the computer program in the memory to implement the steps of the method introduced in any implementation manner of the first aspect.

[0036] Compared with the prior art, this application has the following beneficial effects:

[0037] The present application discloses a method for performing compensation analysis on a motor insurance claim case, comprising: obtaining business data of a target motor insurance claim case; inputting the business data and a target prompt word into a case analysis model, and directly outputting the analysis result of the target motor insurance claim case corresponding to the target prompt word through the analysis of the business data by the case analysis model. In this way, the time spent on detailed analysis of each case by front-line business personnel in the traditional solution is saved, and the analysis result of the motor insurance claim case is directly output through the case analysis model, which greatly improves the work efficiency of compensation analysis of motor insurance claim cases and reduces the work pressure of front-line operators. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0039] Figure 1 A flowchart of a method for performing compensation analysis on a motor vehicle insurance claim case provided in an embodiment of the present application;

[0040] Figure 2 A schematic diagram of analyzing and processing business data using a case analysis model provided in an embodiment of the present application;

[0041] Figure 3 A schematic diagram of the structure of a device for performing compensation analysis on motor vehicle insurance claims cases provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] Auto insurance claims refer to a series of processes and results in which the insurer handles the claim filed by the insured in accordance with the insurance contract after an accident occurs within the scope of insurance liability. In traditional solutions, front-line business personnel need to spend a lot of time sorting out the facts of vehicle claims, combining insurance product conditions and traffic laws and regulations to determine which insurance clauses of the customer in the auto insurance claims case can be compensated, whether the deductible clause is triggered, and what the compensation ratio is.

[0043] With the improvement of living standards, the number of auto insurance claims cases has continued to increase, and front-line operators have long been overwhelmed. How to reduce the work pressure of front-line operators while ensuring efficient compensation analysis of auto insurance claims cases has become one of the technical issues that need to be urgently solved in the field of auto insurance claims technology.

[0044] The present application discloses a method for analyzing the payment of a motor insurance claim case, comprising: obtaining the business data of a target motor insurance claim case; inputting the business data and a target prompt word into a case analysis model, and outputting the analysis result of the target motor insurance claim case corresponding to the target prompt word through the analysis of the business data by the case analysis model. The case analysis model is a model pre-trained based on a large language model and used to output the analysis result of the motor insurance claim case corresponding to the prompt word.

[0045] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0046] Figure 1 A flowchart of a method for analyzing claims for automobile insurance provided in an embodiment of the present application. Figure 1 As shown, the method for analyzing the compensation of a motor vehicle insurance claim case disclosed in the present application includes:

[0047] S101, obtaining business data of a target motor vehicle insurance claim case.

[0048] The target auto insurance claim case is the auto insurance claim case to be analyzed. The business data of the target auto insurance claim case includes but is not limited to: investigation conversation records, insurance data purchased by users, insurance product terms or accident liability.

[0049] In order to obtain more accurate analysis results of the target auto insurance claim case, the target auto insurance claim case selected in this application must include accident liability and insurance data purchased by the user.

[0050] It is understandable that the investigation conversation record can be obtained through manual recording, recording or video recording, etc.; the user insurance product data can be obtained through the official website of the insurance company; the accident liability data can be obtained from the Internet traffic safety comprehensive service management platform, etc. This application does not limit the method of obtaining the business data of the target auto insurance claim case. Those skilled in the art can adopt different methods and channels to obtain more types of business data of the target auto insurance claim case under the premise of legality and compliance.

[0051] It should be noted that the user information (including but not limited to user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0052] S102, obtaining an analysis result of a target motor vehicle insurance claim case corresponding to the target prompt word through a case analysis model according to the business data and the target prompt word.

[0053] The role of the target prompt word is to guide the case analysis model to output the analysis results of the target auto insurance claim case that matches the target prompt word.

[0054] For example, the target prompt word may be: You are an industry expert in accident liability determination, and you need to analyze the input business data and the data such as road traffic regulations that can be obtained, determine the liability of the car insurance claim case, and explain the reasons.

[0055] The target prompt words can be various. Those skilled in the art can set the content of the target prompt words according to actual needs, so as to obtain the analysis results of the target automobile insurance claim cases corresponding to the target prompt words.

[0056] The analysis results include but are not limited to which insurance clauses of the customer in the car insurance claim case can be used for compensation, whether the deductible clause is triggered, and what the compensation ratio is.

[0057] Figure 2 A schematic diagram of a case analysis model provided in an embodiment of the present application for analyzing and processing business data. Figure 2 As shown in the figure, the case analysis model includes an analysis module, an action module, and an evaluation module. Through the coordination of the above three modules, the analysis results of the auto insurance claim case are finally output. The process includes the following steps:

[0058] The first step is to input the business data and target prompt words of the target auto insurance claim case into the case analysis model. The analysis module in the model will analyze the business data and target prompt words to determine whether an analysis result corresponding to the target prompt words can be obtained based on the existing business data.

[0059] If the analysis module believes that the analysis results corresponding to the target prompt words can be obtained based on the existing business data, the analysis results of the target auto insurance claim case will be directly output without activating the action module and the evaluation module; if the analysis module believes that the analysis results corresponding to the target prompt words cannot be obtained based on the existing business data, a tool call plan will be generated.

[0060] The tool calling scheme indicates the types of tools that can be called by the action module in the case analysis model, including but not limited to: a recall tool of RAG (Retrieval-Augmented Generation) or a summary generation tool for auto insurance claims.

[0061] The second step is to input the tool calling plan into the action module after the analysis module outputs the tool calling plan.

[0062] The action module calls the tools in the tool calling plan, that is, through the case analysis model (essentially a large language model), it extracts the factual process of the target auto insurance claim case and obtains the content summary of the target auto insurance claim case; uses RAG to call the recall work to realize the recall of traffic laws and insurance exemption clauses, and finally obtains the call result.

[0063] Among them, the types of content summaries of target auto insurance claims cases are diverse, and can be subdivided into traffic accident judgment summary, accident process summary, loss description summary and accident liability summary.

[0064] The third step is to input the call result into the evaluation module after obtaining the call result.

[0065] The call results are analyzed through the evaluation module to determine whether the call results are related to the review and payment of the motor vehicle insurance claim case, and whether they are helpful for the review and payment of the motor vehicle insurance claim; if they are not helpful or not helpful enough, continue to determine what additional information is needed, and use the results of the above analysis as the judgment result.

[0066] The fourth step is to input the judgment result into the analysis module after obtaining the judgment result, and the analysis module determines whether an analysis result corresponding to the target prompt word can be obtained based on the judgment result, business data and target prompt word.

[0067] If the analysis module believes that, based on these data (judgment results, business data and target prompt words), an analysis result corresponding to the target prompt word can be obtained, then the analysis result of the target motor insurance claim case is directly output.

[0068] If the analysis module believes that the analysis result corresponding to the target prompt word cannot be obtained based on these data (judgment results, business data and target prompt word), the production tool calling plan is updated.

[0069] If an updated production tool calling scheme is obtained, "step one" to "step four" in the aforementioned embodiment are repeatedly executed until the analysis result of the target motor vehicle insurance claim case corresponding to the target prompt word is output.

[0070] From the above analysis, we can see that in the process of analyzing a target auto insurance claim case, the three modules (analysis module, action module and evaluation module) in the case analysis model may be called multiple times. If the process of calling the analysis module, action module and evaluation module once is called a thinking process of the case analysis model, then the analysis of a target auto insurance claim case requires multiple rounds of thinking processes to finally obtain the analysis results of the target auto insurance claim case corresponding to the target prompt word.

[0071] In summary, the present application discloses a method for performing compensation analysis on auto insurance claims cases, including: obtaining business data of a target auto insurance claims case; inputting the business data and target prompt words into a case analysis model, and directly outputting the analysis results of the target auto insurance claims case corresponding to the target prompt words through the analysis of the business data by the case analysis model. In this way, the time spent on detailed analysis of each case by front-line business personnel in the traditional solution is saved, and the analysis results of the auto insurance claims case are directly output through the case analysis model, which greatly improves the work efficiency of compensation analysis of auto insurance claims cases and reduces the work pressure of front-line operators.

[0072] Furthermore, the present application also discloses the specific process of obtaining the case analysis model, including:

[0073] (1) Obtain multiple sets of sample data. Each set of sample data includes: historical business data of historical motor insurance claims cases and analysis results of the historical motor insurance claims cases.

[0074] For the historical business data of historical motor vehicle insurance claims cases, please refer to the relevant introduction about the business data of the target motor vehicle insurance claims cases in the aforementioned embodiment, which will not be repeated here.

[0075] (2) The large language model to be trained is trained based on sample prompt words and multiple groups of sample data to obtain a case analysis model.

[0076] The large language model to be trained used in this application is the qwen1.572B large model that uses React technology. The function of Planning in React technology is equivalent to the function of the analysis module in this application; the function of Action in React technology is equivalent to the function of the action module in this application; the function of Critic in React technology is equivalent to the function of the evaluation module in this application.

[0077] Since the large language model to be trained in this application is the qwen1.572B large model combined with React technology, the large language model to be trained in this application has multi-round thinking capabilities and can output the analysis results of auto insurance claims cases based on data such as survey conversation records, insurance purchased by users, insurance product terms, traffic laws and regulations, etc.

[0078] It should be noted that those skilled in the art may adopt other large language models as the large language model to be trained, and the specific form of the large language model to be trained is not limited in this application.

[0079] Next, combining the historical business data of a specific historical auto insurance claim case, the analysis results of the historical auto insurance claim case and the sample prompt words, the process of training a large language model based on the sample prompt words and sample data to obtain a case analysis model is described in detail.

[0080] For example, the historical business data of historical auto insurance claims cases include: insurance purchased by the user and the inspection conversation record. Specifically:

[0081] The insurance purchased by the user is: "Compulsory Motor Vehicle Traffic Accident Liability Insurance Clauses".

[0082] The contents of the investigation dialogue record are as follows:

[0083] Surveyor: Hello, I am a car insurance surveyor. I will provide you with survey services.

[0084] Customer: Hello, surveyor. Yesterday, I was driving my car and had an accident. A bicycle rode on the motor vehicle lane. My driver's license was cancelled according to law, so I secretly drove on the road. The two vehicles collided directly. Because the road was slippery, I didn't control the brakes well, and the car lost control and hit the guardrail on the side of the road. The right side of the vehicle was seriously damaged, including some scratches on the front bumper, headlights and right side of the body. As for the surrounding environment, it was a small road with two-way single lane. There were no other vehicles on the road at the time, only the lane on my side had water, which might be left by the rain the night before.

[0085] Surveyor: OK, I understand the basic situation. Can you tell me the exact time when the accident happened?

[0086] Customer: The accident happened at three o'clock in the afternoon.

[0087] Surveyor: I see. The accident happened at 3pm. The weather was a bit gloomy and the road was slippery. What is the make and model of your vehicle?

[0088] Customer: My vehicle is a Toyota Camry, the specific model is 2018. I remember there should be detailed records on the insurance policy. The car is about four or five years old and is usually well maintained.

[0089] Exemplarily, the sample prompt words are: generating the damage situation of both vehicles, the responsibility for the accident, and whether it is a traffic accident.

[0090] The sample prompt words, the insurance purchased by the user and the inspection conversation records are input into the large language model to be trained. After multiple rounds of thinking process, the large language model to be trained finally obtains the analysis results corresponding to the sample prompt words.

[0091] The specific process of multiple rounds of thinking is as follows:

[0092] First round of thought process:

[0093] Analysis module: Analyze the input historical business data and sample prompt words (generate the loss of both vehicles, the responsibility for the accident and whether it is a traffic accident), and generate a tool call plan.

[0094] Action module: Call the tool to the summary generation tool for the auto insurance claim case in the solution, and generate the following call results:

[0095] [Traffic Accident Judgment]: Yes, the car owner’s operating error caused the vehicle to collide with the guardrail.

[0096] [Accident Process]: The other party's bicycle was ridden on the motor vehicle lane, and the customer's registration was cancelled according to law but he was still able to drive on the road normally, and the two vehicles collided directly with each other.

[0097] [Description of loss]: Customer loss: There was a loss, the right side of the vehicle was damaged, including the deformed front bumper, broken headlights, scratches on the body, crooked rearview mirror, and scratches on the tires. The other party's loss: There was no loss, because the accident only involved the owner's vehicle. Other losses: There was a loss, and the roadside guardrail was damaged.

[0098] [Accident Responsibility]: Both parties violated traffic regulations, and it is impossible to determine the primary and secondary factors and the specific responsibility for the accident.

[0099] Evaluation module: further analyze the call results to obtain judgment results.

[0100] The action results obtained from the previous action plan include relevant loss information, accident responsibility information, and traffic accident judgment information, which are helpful for the review of auto insurance claims. However, it should be noted that the responsibility of both parties for the accident has not been clearly determined.

[0101] Second round of thought process:

[0102] Analysis module: Based on the call results, use the "Traffic Law Recall Tool" to input the accident process and obtain the relevant road traffic legal basis.

[0103] Action module: Call the "Traffic Law Recall Tool" to generate updated call results.

[0104] Article 57 of Chapter 4, Section 3 of the Road Traffic Safety Law: When driving a non-motor vehicle on a road, relevant traffic safety regulations must be observed. Non-motor vehicles must drive in non-motor vehicle lanes; on roads without non-motor vehicle lanes, they must drive on the right side of the road.

[0105] Article 19 of Chapter 2, Section 2 of the Road Traffic Safety Law: To drive a motor vehicle, one must obtain a motor vehicle driver's license in accordance with the law. To apply for a motor vehicle driver's license, one must meet the driving license conditions stipulated by the public security department. After passing the test, the traffic management department of the public security organ shall issue a motor vehicle driver's license of the corresponding category. A person holding a foreign motor vehicle driver's license who meets the driving license conditions stipulated by the public security department and passes the examination by the traffic management department of the public security organ may be issued a Chinese motor vehicle driver's license. The driver shall drive a motor vehicle in accordance with the type of vehicle permitted by the driver's license; when driving a motor vehicle, the driver must carry the motor vehicle driver's license with him. No unit or individual other than the traffic management department of the public security organ may confiscate or detain a motor vehicle driver's license.

[0106] Evaluation module: further analyzes the call results to obtain updated judgment results.

[0107] The action results obtained from the previous action plan have obtained relevant laws, which are helpful for determining the accident responsibility in the auto insurance claim review. According to the law obtained from the recall, the accident responsibility is determined to be the customer's primary responsibility and the bicycle's secondary responsibility. After determining the accident responsibility, it is also necessary to determine whether the insurance exemption clause is triggered.

[0108] The third round of thinking process:

[0109] Analysis module: Based on the updated judgment results, use the "Insurance Exemption Clause Recall Tool" to input the accident process and obtain the insurance exemption clause.

[0110] Action module: Call the "Insurance Exemption Clause Recall Tool" to generate updated call results.

[0111] Article 23 of the Compulsory Motor Vehicle Traffic Accident Liability Insurance Clauses provides that in the following three circumstances, the insured may request the termination of the compulsory traffic accident liability insurance contract: (1) The insured motor vehicle is deregistered in accordance with the law.

[0112] Evaluation module: Analyzes the updated call results and generates updated judgment results.

[0113] The action result obtained from the previous action plan was the deductible clause of the "Motor Vehicle Traffic Accident Liability Compulsory Insurance Clauses". The relevant deductible clause is helpful for vehicle claims investigation and does not require further relevant information.

[0114] Fourth round of thinking process:

[0115] Analysis module: obtains analysis results based on the updated judgment results, historical business data and sample prompt words obtained in the third round.

[0116] According to the content indexed by the insurance exemption clauses, because the customer violated the "relevant exemption clauses" that caused the insured motor vehicle to be deregistered according to law, the insured can request to terminate the insurance contract. Therefore, no claims will be made for the "Motor Vehicle Traffic Accident Liability Compulsory Insurance Clauses" purchased by the customer.

[0117] The results of the fourth round of analysis module input are compared with the analysis results of the historical auto insurance claim case (the results of manual analysis), and the large language model to be trained is trained so that the model to be trained knows which module should be enabled and which functions should be implemented in each round of thinking, and finally a case analysis model is obtained.

[0118] Since fine-tuning a large language model to be trained based on training data is well known to those skilled in the art, this content will not be further introduced in this application.

[0119] The case analysis model disclosed in this application has the following advantages:

[0120] First, the case analysis model in this application is trained based on the qwen1.572B large model combined with React technology. The model has the ability to think, judge, and reason, and is closer to human judgment.

[0121] Second, the case analysis model in this application is highly reusable and can be applied to claims scenarios of more types of insurance cases after changing the training data.

[0122] Third, compared with traditional deep learning models, large language models have better results, higher generalization ability, better understanding and reasoning ability.

[0123] Fourth, the analysis module, action module and call module are uniformly arranged on a large language model, which has a higher degree of task aggregation and lower maintenance and deployment costs.

[0124] Fifth, the construction of training data is easy, and the large language model to be trained has a large amount of prior knowledge; based on a small amount of data, the case analysis model in this application can be trained.

[0125] Based on the method for performing compensation analysis on automobile insurance claim cases introduced in the aforementioned embodiment, the present application also discloses a device for performing compensation analysis on automobile insurance claim cases.

[0126] Figure 3 A schematic diagram of the structure of a device for analyzing the compensation of a motor vehicle insurance claim provided in an embodiment of the present application. Figure 3 As shown, the device 300 for analyzing the compensation of a motor vehicle insurance claim case includes:

[0127] The data acquisition module 301 is used to acquire the business data of the target automobile insurance claim case;

[0128] The analysis result acquisition module 302 is used to obtain the analysis results of the target motor vehicle insurance claim case corresponding to the target prompt word through the case analysis model according to the business data and the target prompt word; the case analysis model is a model pre-trained based on the large language model and is used to output the analysis results of the motor vehicle insurance claim case corresponding to the prompt word.

[0129] In an optional implementation, the analysis result acquisition module 302 includes:

[0130] A data analysis unit, configured to generate, through the analysis module, any one of a tool calling scheme and a preliminary analysis result according to the business data and the target prompt word;

[0131] A first analysis result obtaining unit, configured to use the preliminary analysis result as the analysis result if the preliminary analysis result is output;

[0132] The second analysis result acquisition unit is used to generate the analysis result according to the business data, the target prompt word and the tool calling plan through the action module, the evaluation module and the analysis module if the tool calling plan is output.

[0133] In an optional implementation, the second analysis result obtaining unit includes:

[0134] A call result acquisition subunit is used to call the tool in the tool call scheme through the action module to obtain a call result; the call result includes: a summary of traffic laws, insurance clauses or the target auto insurance claim case;

[0135] A judgment result acquisition subunit is used to analyze the call result through the evaluation model to obtain a judgment result; the judgment result includes whether it is necessary to obtain supplementary information for analyzing the target motor insurance claim case, and the name of the supplementary information;

[0136] The second analysis result acquisition subunit is used to generate an updated tool calling scheme or the analysis result through the analysis module based on the judgment result, the business data and the target prompt word; the updated tool calling scheme is used as input data of the evaluation module to generate the analysis result.

[0137] In an optional implementation, the apparatus 300 for performing compensation analysis on a motor vehicle insurance claim case further includes:

[0138] A sample data acquisition module is used to acquire multiple groups of sample data, each group of sample data includes: historical business data of historical motor insurance claims cases and analysis results of the historical motor insurance claims cases;

[0139] The case analysis model training module is used to train the large language model to be trained based on the sample prompt words and multiple groups of sample data to obtain the case analysis model.

[0140] Based on the method and device for performing compensation analysis on motor vehicle insurance claims cases provided in the aforementioned embodiments, the present application also provides a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, implements some or all of the steps in the method for performing compensation analysis on motor vehicle insurance claims cases mentioned above.

[0141] Based on the method and device for analyzing the compensation of automobile insurance claims provided in the above embodiments, the present application further provides an electronic device, including:

[0142] a memory having a computer program stored thereon;

[0143] A processor is used to execute the computer program in the memory to implement some or all of the steps in the method for performing compensation analysis on motor insurance claims cases provided in the above-mentioned embodiment.

[0144] It should be noted that each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is merely schematic, in which the unit described as a separate component may or may not be physically separated, and the component prompted as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative work.

[0145] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for analyzing the payment of automobile insurance claims, characterized in that: The method comprises: Obtain business data of target auto insurance claims cases; Through the case analysis model, according to the business data and the target prompt word, the analysis results of the target motor insurance claim case corresponding to the target prompt word are obtained; the case analysis model is a model pre-trained based on a large language model, and is used to output the analysis results of the motor insurance claim case corresponding to the prompt word.

2. The method according to claim 1, characterized in that The case analysis model includes an analysis module, an action module and an evaluation module; the case analysis model obtains the analysis results of the target auto insurance claim case corresponding to the target prompt word according to the business data and the target prompt word, including: Generate, by the analysis module, either a tool calling scheme or a preliminary analysis result according to the business data and the target prompt word; If the preliminary analysis result is output, the preliminary analysis result is used as the analysis result; If the tool calling scheme is output, the analysis result is generated according to the business data, the target prompt word and the tool calling scheme through the action module, the evaluation module and the analysis module.

3. The method according to claim 2, characterized in that The generating of the analysis result according to the business data, the target prompt word and the tool calling scheme through the action module, the evaluation module and the analysis module includes: Through the action module, the tool in the tool calling scheme is called to obtain a calling result; the calling result includes: a summary of traffic laws, insurance clauses or the target auto insurance claim case; By using the evaluation model, the call result is analyzed to obtain a judgment result; the judgment result includes whether it is necessary to obtain supplementary information for analyzing the target auto insurance claim case, and the name of the supplementary information; Through the analysis module, an updated tool calling scheme or the analysis result is generated based on the judgment result, the business data and the target prompt word; the updated tool calling scheme is used as input data of the evaluation module to generate the analysis result.

4. The method according to claim 1, characterized in that: The training steps of the case analysis model include: Acquire multiple groups of sample data, each group of sample data includes: historical business data of historical motor insurance claims cases and analysis results of the historical motor insurance claims cases; The large language model to be trained is trained based on the sample prompt words and the multiple groups of sample data to obtain the case analysis model.

5. The method according to any one of claims 1 to 4, characterized in that The business data includes: survey conversation records, insurance data purchased by users, insurance product terms or accident liability.

6. A device for analyzing the payment of automobile insurance claims, characterized in that: include: Data acquisition module, used to obtain business data of target auto insurance claims cases; An analysis result acquisition module, used to obtain the analysis result of the target motor vehicle insurance claim case corresponding to the target prompt word through the case analysis model according to the business data and the target prompt word; The case analysis model is a model that is pre-trained based on a large language model and is used to output analysis results of motor vehicle insurance claims cases corresponding to prompt words.

7. The device according to claim 6, characterized in that The analysis result acquisition module includes: A data analysis unit, configured to generate, through the analysis module, any one of a tool calling scheme and a preliminary analysis result according to the business data and the target prompt word; A first analysis result obtaining unit, configured to use the preliminary analysis result as the analysis result if the preliminary analysis result is output; The second analysis result acquisition unit is used to generate the analysis result according to the business data, the target prompt word and the tool calling plan through the action module, the evaluation module and the analysis module if the tool calling plan is output.

8. The device according to claim 7, characterized in that The second analysis result obtaining unit includes: A call result acquisition subunit is used to call the tool in the tool call scheme through the action module to obtain a call result; the call result includes: a summary of traffic laws, insurance clauses or the target auto insurance claim case; A judgment result acquisition subunit is used to analyze the call result through the evaluation model to obtain a judgment result; the judgment result includes whether it is necessary to obtain supplementary information for analyzing the target motor insurance claim case, and the name of the supplementary information; The second analysis result acquisition subunit is used to generate an updated tool calling scheme or the analysis result through the analysis module based on the judgment result, the business data and the target prompt word; the updated tool calling scheme is used as input data of the evaluation module to generate the analysis result.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 5 are implemented.

10. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 5.

Images