Case report information processing and geographic position pushing method based on similar case matching

Abstract

The invention provides a case report information processing and geographic position pushing method based on similar case matching. Relates to the technical field of insurance case processing. The target claim settlement case similar to the current case is matched from the insurance case library in a similar case matching mode, and the process file of the target claim settlement case is provided to construct the claim settlement task of the current case. According to the method, case reporting data and insurance rule provisions are associated by utilizing comparative learning, and case-related legal provisions are configured in the case reporting data. According to the method, similarity matching of the claim settlement cases is carried out from three aspects of claim settlement case events, laws related to the claim settlement cases and insurance rules related to the claim settlement cases. The matching result considers the semantics of the text describing the event, and also considers laws and insurance rules behind the event, so that the matched target claim settlement case is more accurate, and the claim settlement task constructed based on the process file of the target claim settlement case better meets the requirements of the current claim settlement case.

Description

Technical Field

[0001] The present invention relates to the technical field of insurance case processing, and in particular to a method for processing case report information and pushing geographic location based on similar case matching. Background Art

[0002] With the rapid development of information technology, intelligent services have penetrated into all areas of society. In the fields of public security and financial insurance, how to quickly and accurately respond to and process user reports, improve the efficiency and satisfaction of claims services, make good use of claims customer resources, and realize the redevelopment of claims customer resources has always been the goal.

[0003] Since the traditional insurance claims process relies on manual processing, it takes time to input, review and schedule information, and the response time is long and inefficient, making it difficult to achieve emergency response, resulting in a decline in customer satisfaction. At the same time, a large amount of historical claims data has not been fully utilized, relying on expert experience, and it is difficult to extract valuable information from a large amount of data according to actual conditions, and it is impossible to provide support for real-time decision-making. With the development of natural language processing technology, similar case matching based on text similarity analysis is used in the insurance case processing process. By matching similar cases, similar cases can be found in historical claims cases, and the claims process of historical claims cases can be used to assist the current case. However, the existing case similarity matching is mostly based on the analysis of semantic similarity from the perspective of event facts, which is limited to the analysis of semantic similarity. It is difficult to take into account the deep legal and insurance rules hidden behind the event. Legal information and insurance rules are conditions that cannot be ignored in the claims process, which makes the range of historical cases matched too large and the effect of assisting claims is poor. Summary of the invention

[0004] In order to solve the above technical problem or at least partially solve the above technical problem, the present invention provides a method for processing report information and pushing geographic location based on similar case matching.

[0005] In a first aspect, the present invention provides a method for processing case report information and pushing geographic location information based on similar case matching, comprising:

[0006] For different types of insurance, a corresponding insurance case library is constructed, wherein the insurance case library stores insurance claim cases and process files corresponding to each claim case; the claim cases in the insurance case library are added with preset legal and insurance rule tags at the sentence level;

[0007] Collect the report data of the claim settlement customer and configure the corresponding report time; the report data includes the reporter information, insurance type, geographical location information of the incident location, incident time and case description; the case description includes the relationship between the persons involved, the detailed process of the case, the determination of the case responsibility, and the legal provisions involved;

[0008] Matching the case report data with the corresponding insurance rules and regulations within the scope of the insurance rule base by comparing the second text encoder and the first text encoder trained by learning;

[0009] Label the legal provisions and insurance rules in the case report data;

[0010] Using prompt words to control the language model fine-tuned in advance for similar case matching tasks, matching the target claim case with the largest similarity score from the insurance case database according to the case report data labeled with legal provisions and insurance rules, and using the process file of the screened target claim case to build the claim task of the current case;

[0011] The geographical information of the claim settlement task of the current case is displayed based on the geographical location information of the crime scene.

[0012] Furthermore, the process of adding preset legal and insurance rules labels to claim cases at the sentence level includes:

[0013] According to different types of insurance, organize the laws and insurance rules in the corresponding fields; represent the provisions in the laws and insurance rules with different numbers, and form different lists corresponding to the numbers of the provisions in the laws and insurance rules as legal labels and insurance rule labels;

[0014] For claim cases, the content is analyzed sentence by sentence, and corresponding legal labels and insurance rule label annotations are added according to the legal provisions or insurance rule provisions involved in each sentence, so that the case facts in the claim case contain in-depth legal and insurance rule information.

[0015] Furthermore, after obtaining the case report data, the case report data of the claim customer is structured according to the claim case format in the insurance case library.

[0016] Furthermore, the second text encoder and the first text encoder are trained by contrastive learning, and the process includes:

[0017] Construct contrastive learning training set;

[0018] The second text encoder is constructed based on the Lawbert model fine-tuned for insurance rules, and the first text encoder is constructed based on the Bert model; the last Transformer layer of the Lawbert model and the Bert model is used as the part with adjustable parameters in the comparative learning process, and the parameters of the remaining Transformer layers are fixed;

[0019] The second text encoder is used to extract the text features of any claim case sentence, and the first text encoder is used to extract the text features of the insurance rules and regulations in the positive and negative pairs of the claim case sentence in the contrastive learning training set:

[0020] Text features of case sentences:

[0021] F 1d =Enc1(Case_s d ), where Enc1() is the first text encoder, Case_s d For the claim case sentence;

[0022] Text features of insurance rules in the claims case sentences:

[0023] F 2d =Enc2(Insurance_Rules d ), where Enc2() is the second text encoder, Insurance_Rules d Case_s for claim cases d The insurance rules and regulations are directly in place;

[0024] Text features of the insurance rules in all negative pairs of case sentences:

[0025] F 2k =Enc2(Insurance_Rules k ),k=1,2...D,k≠d, where, Insurance_Rules k Case_s for claim cases d The kth negative pair insurance rule clause;

[0026] The objective function in the contrastive learning process is as follows:

[0027]

[0028] Where D is the total number of positive and negative pairs of case sentences, cossim() is the function for calculating cosine similarity, exp() is the exponential function of the natural logarithm, and τ is the learnable temperature coefficient;

[0029] The objective function is used to constrain the adjustment of the parameters of the last Transformer layer of the second text encoder and the first text encoder so that within the scope of the contrastive learning training set, the similarity of text features extracted from the positive pairs by the second text encoder and the first text encoder is maximized, and the similarity of text features extracted from the negative pairs is minimized.

[0030] Furthermore, according to the correspondence between the cases in the insurance case library and the insurance rule labels in the cases, the matching relationship between the sentences in the cases and the insurance rule clauses is determined, and the matching case sentences and insurance rule clauses form positive training data for contrastive learning; the insurance rules are randomly changed so that the case sentences and the insurance rule clauses no longer match, thereby obtaining negative training data for contrastive learning; the case sentences are expanded using the GPT model while ensuring that the semantics remain unchanged; the corresponding insurance rules after the expansion are still the insurance rules corresponding to the original sentences, thereby obtaining expanded positive training data and negative training data.

[0031] Furthermore, the process of fine-tuning the language model includes:

[0032] Randomly select some of the various claim cases in the insurance case database as matching case sources, manually select the claim cases most relevant to the matching case sources from the remaining claim cases as matching results, and construct a matching training set using the matching case sources and the matching results;

[0033] The language model is trained to select predicted matching results from the remaining claim cases based on the matching case sources, and the language model parameters are fine-tuned based on the information cross entropy loss between the predicted matching results and the actual matching results in the matching training set as the loss function.

[0034] Furthermore, the trained second text encoder and the first text encoder are packaged into a Docker image, and the Docker image is deployed as an API service for configuring insurance rules and regulations. By calling the deployed API service of insurance rules and regulations, the corresponding insurance rules and regulations involved in the case are matched for the report data; the trained language model is packaged into a Docker image, and the language model is deployed as a claim case matching API service. By calling the deployed claim case matching API service, the current claim case is processed in real time to build a claim task.

[0035] Furthermore, the geographical information display of the current case's claim settlement task is realized based on the geographical location information of the crime scene, including: the front end obtains the geographical location information of the crime scene in the report data from the back end, obtains the report timestamp, and obtains the reporter information;

[0036] Map visualization based on the geographical location of the crime scene:

[0037] Set up a claim task icon at the location of the visual map;

[0038] The first operation on the claim task icon triggers navigation, navigating from the current location to the crime scene corresponding to the claim task;

[0039] In the visual map, a description card of the claim task that can be displayed or hidden is configured. The second operation on the claim task icon triggers the display or hiding of the description card. The description card is designed through the UI to provide the reporting time, reporter information, and links to obtain case status and process files. An information export button is set in the description card. Click the information export button to obtain the reporting time, crime scene, reporter information, and case status and process files of the claim task.

[0040] In the second aspect, the present invention provides a device for processing report information and pushing geographic location based on similar case matching, including: back-end and front-end devices connected to each other, the back-end and front-end devices include at least one processing unit, the processing unit and the storage unit are interconnected through a bus unit, the storage unit stores a computer program, and when the computer program is executed by the processing unit, the method for processing report information and pushing geographic location based on similar case matching is implemented.

[0041] In a third aspect, the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the method for processing report information and pushing geographic location based on similar case matching.

[0042] The above technical solution provided by the embodiment of the present invention has the following advantages compared with the prior art:

[0043] This application will match a target claim case similar to the current case from the insurance case library by means of similar case matching, and provide a process file of the target claim case to construct the claim task of the current case. The present invention performs similarity matching of claim cases from three aspects: claim case events, laws involved in claim cases, and insurance rules involved in claim cases. The matching result not only considers the semantics of the text describing the event, but also the laws and insurance rules behind the event, so that the matched target claim case is more accurate, and the claim task constructed based on the process file of the target claim case is more in line with the requirements of the current claim case.

[0044] This application realizes the geographical information display of the claim task of the current case based on the geographical location information of the crime scene, and more intuitively displays the claim task status and processing results on the front end. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0046] In order to more clearly illustrate the embodiments of the present invention 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, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0047] Figure 1 A flowchart of a method for processing case report information and pushing geographic location based on similar case matching provided by an embodiment of the present invention;

[0048] Figure 2 A flowchart of adding preset legal and insurance rule labels to a claim case at the sentence level provided by an embodiment of the present invention;

[0049] Figure 3 A flowchart of training a second text encoder and a first text encoder by contrastive learning provided by an embodiment of the present invention;

[0050] Figure 4 A flowchart of fine-tuning a language model provided in an embodiment of the present invention so that the language model can better adapt to the task of matching similar claim cases;

[0051] Figure 5 A schematic diagram of a device for processing report information and pushing geographic location based on similar case matching provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0054] Example 1

[0055] like Figure 1 As shown, the technology of the present invention realizes a method for processing case report information and pushing geographic location based on similar case matching. The present invention adopts a design architecture with front-end and back-end separation. The back-end is responsible for collecting case report data and matching similar cases, and constructs the claim settlement task of the current case based on the claim settlement process file of the matched similar cases. The front-end performs GIS display of the claim settlement task of the current case based on the geographic location information of the crime scene.

[0056] The process of the present invention comprises:

[0057] The insurance case library is configured at the back end. This application will match the target claim case similar to the current case from the insurance case library by similar case matching, and provide the process file of the target claim case to construct the claim task of the current case.

[0058] In the specific implementation process, for different types of insurance, a corresponding insurance case library is constructed. The insurance case library stores insurance claim cases and process files corresponding to each claim case. The claim cases in the insurance case library are added with preset legal and insurance rule tags at the sentence level.

[0059] like Figure 2 As shown in Figure 1, the process of adding preset legal and insurance rule labels to claim cases at the sentence level includes:

[0060] Define the legal and insurance rule label formats in advance based on the laws and insurance rules. Specifically, organize the laws and insurance rules in the corresponding fields for different types of insurance. Use different numbers to represent the articles in the laws and insurance rules, and form different lists corresponding to the numbers of the articles in the laws and insurance rules as legal labels and insurance rule labels.

[0061] For claim cases, the content is analyzed sentence by sentence, and corresponding legal labels and insurance rule label annotations are added according to the legal provisions or insurance rule provisions involved in each sentence, so that the case facts in the claim case contain in-depth legal and insurance rule information.

[0062] Collect the report data of claim customers and configure the corresponding report time; the report data includes the reporter information, insurance type, geographical location information of the incident, incident time and case description. The case description includes the relationship between the persons involved, the detailed process of the case, the determination of the case responsibility, and the legal provisions involved. The legal provisions involved are given by the public security and court law enforcement departments.

[0063] During the implementation process, the claim customer's report data is captured through the insurance call center API or the insurance report system API. When the claim customer successfully submits the report data, a timestamp representing the report time is configured for the report data based on the insurance call center API or the insurance report system API local time. During the specific implementation process, in order to improve the claim case matching effect, the claim customer's report data is structured according to the claim case format in the insurance case library.

[0064] The second text encoder and the first text encoder are deployed on the back end, and the insurance rule base is deployed on the back end. The second text encoder and the first text encoder trained by comparative learning are used within the scope of the insurance rule base to match the corresponding insurance rules and regulations for the case report data. The trained second text encoder and the first text encoder are packaged into a Docker image, and the Docker image is deployed as an API service for configuring insurance rules and regulations. By calling the API service of the deployed insurance rules and regulations, the corresponding insurance rules and regulations for the case report data are matched.

[0065] In the specific implementation process, Figure 3 As shown, the second text encoder and the first text encoder are trained by contrastive learning, and the process includes:

[0066] Construct a contrastive learning training set; determine the matching relationship between the sentences in the case and the insurance rule clauses according to the corresponding relationship between the cases in the insurance case library and the insurance rule labels in the cases, and the case sentences and insurance rule clauses that match each other form the positive training data for contrastive learning; randomly change the insurance rules so that the case sentences and insurance rule clauses no longer match, and obtain the negative training data for contrastive learning. Use the gpt model to expand the case sentences while ensuring that the semantics remain unchanged. The corresponding insurance rules after expansion are still the insurance rules corresponding to the original sentences. Thus, the expanded positive training data and negative training data are obtained.

[0067] The second text encoder is built based on the Lawbert model fine-tuned for insurance rules, and the first text encoder is built based on the Bert model; the last Transformer layer of the Lawbert model and the Bert model is used as the part with adjustable parameters in the comparative learning process, and the parameters of the remaining Transformer layers are fixed.

[0068] The second text encoder is used to extract the text features of any case sentence, and the first text encoder is used to extract the text features of the insurance rules and regulations in the positive and negative pairs of the case sentence in the contrastive learning training set:

[0069] Text features of case sentences:

[0070] F 1d =Enc1(Case_s d ), where Enc1() is the first text encoder, Case_s d For the claim case sentence;

[0071] Text features of insurance rules in the claims case sentences:

[0072] F 2d =Enc2(Insurance_Rules d ), where Enc2() is the second text encoder, Insurance_Rules d Case_s for claim cases d The insurance rules and regulations are directly in place;

[0073] Text features of the insurance rules in all negative pairs of case sentences:

[0074] F 2k =Enc2(Insurance_Rules k ),k=1,2...D,k≠d, where, Insurance_Rules k Case_s for claim cases d The kth negative pair insurance rule clause;

[0075] The objective function in the contrastive learning process is as follows:

[0076]

[0077] Where D is the total number of positive and negative pairs of case sentences, cossim() is the function for calculating cosine similarity, exp() is the exponential function of the natural logarithm, and τ is the learnable temperature coefficient.

[0078] The objective function is used to constrain the adjustment of the parameters of the last Transformer layer of the second text encoder and the first text encoder so that within the scope of the contrastive learning training set, the similarity of text features extracted from the positive pairs by the second text encoder and the first text encoder is maximized, and the similarity of text features extracted from the negative pairs is minimized.

[0079] When the trained second text encoder and the first text encoder are deployed at the back end to match the case report data with the corresponding insurance rules and clauses involved in the case, the insurance rules and clauses are selected within the scope of the insurance rule library.

[0080] Label the legal provisions and insurance rules in the report data; in the specific implementation process, label the legal provisions and insurance rules in the insurance data according to the construction method of the legal labels and insurance rules labels of this application.

[0081] The prompt word control language model is used to match the target claim case with the largest similarity score from the insurance case database according to the case report data labeled with legal provisions and insurance rules.

[0082] During the specific implementation process, the language model is fine-tuned to make it better adapt to the task of matching similar claims cases, such as Figure 4 As shown, the process includes:

[0083] Randomly select some of the various claim cases in the insurance case database as matching case sources; hide the insurance rule labels and legal labels of the matching case sources, and manually select the claim cases most relevant to the matching case sources from the remaining claim cases as matching results. Use the matching case sources and matching results to build a matching training set.

[0084] The training language model selects the predicted matching results from the remaining claim cases according to the matching case sources. The information cross entropy loss between the predicted matching results and the actual matching results in the matching training set is used as the loss function to fine-tune the language model parameters to better match similar claim cases.

[0085] Use the process documents of the screened target claims cases to build the claims task of the current case.

[0086] During the specific implementation process, the trained language model is packaged into a Docker image, and the language model is deployed as a claim case matching API service. By calling the deployed claim case matching API service, the current claim case is processed in real time to build a claim task.

[0087] The geographical information display of the current case's claim settlement task is realized based on the geographical location information of the crime scene, including: the front end obtains the geographical location information of the crime scene in the report data from the back end, obtains the report timestamp, and obtains the reporter information.

[0088] Map visualization is performed based on the geographical location information of the crime scene, including:

[0089] Set the claim task icon at the map issuance location on the visualization map.

[0090] Navigation is triggered by the first operation on the claim task icon, and navigation is performed from the current location to the crime scene corresponding to the claim task.

[0091] In the visual map, a description card of the claim task that can be displayed or hidden is configured. The second operation on the claim task icon triggers the display or hiding of the description card. The description card is designed through the UI to provide the reporting time, reporter information, and links to obtain case status and process files. An information export button is set in the description card. Click the information export button to obtain the reporting time, crime scene, reporter information, and case status and process files of the claim task.

[0092] In the specific implementation process, the reporting time is graded, such as: reporting within 10 minutes belongs to the first level, reporting within 20 minutes belongs to the second level, and so on. The reporting time is calculated based on the reporting time of the claim task and the current time, and different borders or background colors are configured for the description card according to the level of the reporting time, so as to more clearly distinguish the importance of the claim tasks.

[0093] Example 2

[0094] See also Figure 5 As shown, an embodiment of the present invention provides a device for processing report information and pushing geographic location based on similar case matching, including: interconnected back-end and front-end devices, the back-end and front-end devices include at least one processing unit, the processing unit and the storage unit are interconnected through a bus unit, and the storage unit is a computer-readable storage medium that can be used to store software programs, computer executable programs and modules, such as the software programs, computer executable programs and modules corresponding to the method for processing report information and pushing geographic location based on similar case matching in the embodiment of the present invention. The processing unit implements the above-mentioned method for processing report information and pushing geographic location based on similar case matching by running the software programs, computer executable programs and modules stored in the storage unit, including:

[0095] For different types of insurance, a corresponding insurance case library is constructed, wherein the insurance case library stores insurance claim cases and process files corresponding to each claim case; the claim cases in the insurance case library are added with preset legal and insurance rule tags at the sentence level;

[0096] Collect the report data of the claim settlement customer and configure the corresponding report time; the report data includes the reporter information, insurance type, geographical location information of the incident location, incident time and case description; the case description includes the relationship between the persons involved, the detailed process of the case, the determination of the case responsibility, and the legal provisions involved;

[0097] Matching the case report data with the corresponding insurance rules and regulations within the scope of the insurance rule base by comparing the second text encoder and the first text encoder trained by learning;

[0098] Label the legal provisions and insurance rules in the case report data;

[0099] Using prompt words to control the language model fine-tuned in advance for similar case matching tasks, matching the target claim case with the largest similarity score from the insurance case database according to the case report data labeled with legal provisions and insurance rules, and using the process file of the screened target claim case to build the claim task of the current case;

[0100] The geographical information of the claim settlement task of the current case is displayed based on the geographical location information of the crime scene.

[0101] Of course, the storage unit in the device for processing report information and pushing geographic location based on similar case matching provided in an embodiment of the present invention, the computer program stored therein is not limited to the operation of the method described above, and can also execute the relevant operations in the method for processing report information and pushing geographic location based on similar case matching provided in any embodiment of the present invention.

[0102] Example 3

[0103] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed, the method for processing case report information and pushing geographic location based on similar case matching is implemented, including:

[0104] For different types of insurance, a corresponding insurance case library is constructed, wherein the insurance case library stores insurance claim cases and process files corresponding to each claim case; the claim cases in the insurance case library are added with preset legal and insurance rule tags at the sentence level;

[0105] Collect the report data of the claim settlement customer and configure the corresponding report time; the report data includes the reporter information, insurance type, geographical location information of the incident location, incident time and case description; the case description includes the relationship between the persons involved, the detailed process of the case, the determination of the case responsibility, and the legal provisions involved;

[0106] Matching the case report data with the corresponding insurance rules and regulations within the scope of the insurance rule base by comparing the second text encoder and the first text encoder trained by learning;

[0107] Label the legal provisions and insurance rules in the case report data;

[0108] Using prompt words to control the language model fine-tuned in advance for similar case matching tasks, matching the target claim case with the largest similarity score from the insurance case database according to the case report data labeled with legal provisions and insurance rules, and using the process file of the screened target claim case to build the claim task of the current case;

[0109] The geographical information of the claim settlement task of the current case is displayed based on the geographical location information of the crime scene.

[0110] A computer-readable storage medium provided in an embodiment of the present invention stores a computer program which is not limited to the method operations described above, but can also execute related operations in a method for processing report information and pushing geographic location based on similar case matching provided in any embodiment of the present invention.

[0111] In the embodiments provided by the present invention, it should be understood that the disclosed structures and methods can be implemented in other ways. For example, the structural embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, structures or units, which can be electrical, mechanical or other forms.

[0112] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0113] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0114] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for processing case report information and pushing geographic location based on similar case matching, characterized in that: include: For different types of insurance, a corresponding insurance case library is constructed, wherein the insurance case library stores insurance claim cases and process files corresponding to each claim case; Claim cases in the insurance case database are added with preset legal and insurance rule tags at the sentence level; Collect the report data of the claim settlement customer and configure the corresponding report time; the report data includes the reporter information, insurance type, geographical location information of the incident location, incident time and case description; the case description includes the relationship between the persons involved, the detailed process of the case, the determination of the case responsibility, and the legal provisions involved; Matching the case report data with the corresponding insurance rules and regulations within the scope of the insurance rule base by comparing the second text encoder and the first text encoder trained by learning; Label the legal provisions and insurance rules in the case report data; Using prompt words to control the language model fine-tuned in advance for similar case matching tasks, matching the target claim case with the largest similarity score from the insurance case database according to the case report data labeled with legal provisions and insurance rules, and using the process file of the screened target claim case to build the claim task of the current case; The geographical information of the claim settlement task of the current case is displayed based on the geographical location information of the crime scene.

2. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: The process of adding pre-set legal and insurance rules labels to claim cases at the sentence level includes: According to different types of insurance, organize the laws and insurance rules in the corresponding fields; represent the provisions in the laws and insurance rules with different numbers, and form different lists corresponding to the numbers of the provisions in the laws and insurance rules as legal labels and insurance rule labels; For claim cases, the content is analyzed sentence by sentence, and corresponding legal labels and insurance rule label annotations are added according to the legal provisions or insurance rule provisions involved in each sentence, so that the case facts in the claim case contain in-depth legal and insurance rule information.

3. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: After obtaining the case report data, the case report data of the claim customer is structured according to the claim case format in the insurance case library.

4. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: The second text encoder and the first text encoder are trained by contrastive learning, and the process includes: Construct contrastive learning training set; The second text encoder is constructed based on the Lawbert model fine-tuned for insurance rules, and the first text encoder is constructed based on the Bert model; the last Transformer layer of the Lawbert model and the Bert model is used as the part with adjustable parameters in the comparative learning process, and the parameters of the remaining Transformer layers are fixed; The second text encoder is used to extract the text features of any claim case sentence, and the first text encoder is used to extract the text features of the insurance rules and regulations in the positive and negative pairs of the claim case sentence in the contrastive learning training set: Text features of case sentences: F 1d =Enc1(Case_s d ), where Enc1() is the first text encoder, Case_s d For the claim case sentence; Text features of insurance rules in the claims case sentences: F 2d =Enc2(Insurance_Rules d ), where Enc2() is the second text encoder, Insurance_Rules d Case_s for claim cases d The insurance rules and regulations are directly in place; Text features of the insurance rules in all negative pairs of case sentences: F 2k =Enc2(Insurance_Rules k ),k=1,2...D,k≠d, where, Insurance_Rules k Case_s for claim cases d The kth negative pair insurance rule clause; The objective function in the contrastive learning process is as follows: Where D is the total number of positive and negative pairs of case sentences, cossim() is the function for calculating cosine similarity, exp() is the exponential function of natural logarithm, and τ is the learnable temperature coefficient; The objective function is used to constrain the adjustment of the parameters of the last Transformer layer of the second text encoder and the first text encoder so that within the scope of the contrastive learning training set, the similarity of text features extracted from the positive pairs by the second text encoder and the first text encoder is maximized, and the similarity of text features extracted from the negative pairs is minimized.

5. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: According to the correspondence between the cases in the insurance case library and the insurance rule labels in the cases, the matching relationship between the sentences in the cases and the insurance rule clauses is determined, and the case sentences and insurance rule clauses that match each other form the positive training data for contrastive learning; the insurance rules are randomly changed so that the case sentences and the insurance rule clauses no longer match, thereby obtaining the negative training data for contrastive learning; the case sentences are expanded using the GPT model while ensuring that the semantics remain unchanged; the corresponding insurance rules after expansion are still the insurance rules corresponding to the original sentences, thereby obtaining expanded positive training data and negative training data.

6. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: The process of fine-tuning a language model involves: Randomly select some of the various claim cases in the insurance case database as matching case sources, manually select the claim cases most relevant to the matching case sources from the remaining claim cases as matching results, and construct a matching training set using the matching case sources and the matching results; The language model is trained to select predicted matching results from the remaining claim cases based on the matching case sources, and the language model parameters are fine-tuned based on the information cross entropy loss between the predicted matching results and the actual matching results in the matching training set as the loss function.

7. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: The trained second text encoder and the first text encoder are packaged into a Docker image, and the Docker image is deployed as an API service for configuring insurance rules and regulations. By calling the deployed API service of insurance rules and regulations, the corresponding insurance rules and regulations involved in the case are matched for the report data; the trained language model is packaged into a Docker image, and the language model is deployed as a claim case matching API service. By calling the deployed claim case matching API service, the current claim case is processed in real time to build a claim task.

8. The method for processing case report information and pushing geographic location based on similar case matching according to claim 1, characterized in that: Based on the geographical location information of the crime scene, the geographical information display of the current case's claim settlement task is realized, including: obtaining the geographical location information of the crime scene in the report data from the backend, obtaining the report timestamp, and obtaining the reporter information; Map visualization based on the geographical location of the crime scene: Set up a claim task icon at the location of the visual map; The first operation on the claim task icon triggers navigation, navigating from the current location to the crime scene corresponding to the claim task; In the visual map, a description card of the claim task that can be displayed or hidden is configured. The second operation on the claim task icon triggers the display or hiding of the description card. The description card is designed through the UI to provide the reporting time, reporter information, and links to obtain case status and process files. An information export button is set in the description card. Click the information export button to obtain the reporting time, crime scene, reporter information, and case status and process files of the claim task.

9. A device for processing case report information and pushing geographic location based on similar case matching, characterized in that: include: The back-end and front-end devices include at least one processing unit, the processing unit and the storage unit are interconnected through a bus unit, the storage unit stores a computer program, and when the computer program is executed by the processing unit, the urology question-and-answer method based on clinical data and magnetic resonance imaging analysis as described in any one of claims 1-8 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for processing case report information and pushing geographic location based on similar case matching as described in any one of claims 1 to 8 is implemented.

Images