A case analysis method and device, a storage medium and an electronic device

Abstract

The specification discloses a case analysis method and device, a storage medium and an electronic equipment, wherein the method comprises: obtaining target case data corresponding to a target case type, inputting the target case data into a pre-trained feature extraction model, extracting features of the target case data by using the feature extraction model to obtain target features of the target case data, obtaining first historical data corresponding to the target features in a data center associated with the target case type, extracting features from the first historical data based on the target features to obtain first feature data, obtaining evaluation data based on the first feature data and the target case data, generating a first evaluation report based on the evaluation data, and using the embodiments of the specification, the features extracted based on the target case data are used to obtain historical data, and an evaluation report corresponding to the target case type is generated, thereby reducing the time required for analyzing cases and writing reports, improving the processing efficiency of case analysis, and ensuring the efficiency and accuracy of case analysis.

Description

TECHNICAL FIELD

[0001] The present specification relates to the technical field of computer, and particularly relates to a case analysis method and device, a storage medium and an electronic equipment. BACKGROUND

[0002] Nowadays, in order to prevent the occurrence of cases, it is necessary to analyze and summarize the occurred cases to reduce the probability of occurrence of cases. However, in the actual process of case analysis, there are problems of low efficiency and insufficient analysis ability. SUMMARY

[0003] The present specification provides a case analysis method and device, a storage medium and an electronic equipment, which can obtain historical data based on the features extracted from target case data, generate an evaluation report corresponding to the target case type, thereby reducing the time required for analyzing cases and writing reports, improving the processing efficiency of case analysis, and ensuring the efficiency and accuracy of case analysis.

[0004] In a first aspect, an embodiment of the present specification provides a case analysis method, which comprises:

[0005] obtaining target case data corresponding to a target case type, and inputting the target case data into a pre-trained feature extraction model;

[0006] extracting features of the target case data by using the feature extraction model to obtain target features of the target case data;

[0007] obtaining first historical data corresponding to the target features in a data center associated with the target case type;

[0008] extracting features of the first historical data based on the target features to obtain first feature data;

[0009] obtaining evaluation data based on the first feature data and the target case data, and generating a first evaluation report based on the evaluation data.

[0010] In a second aspect, an embodiment of the present specification provides a case analysis device, which comprises:

[0011] a case data obtaining unit configured to obtain target case data corresponding to a target case type, and input the target case data into a pre-trained feature extraction model;

[0012] a feature extraction unit configured to extract features of the target case data by using the feature extraction model to obtain target features of the target case data;

[0013] a historical data acquisition unit, configured to acquire first historical data corresponding to the target feature from a data center associated with the target case type;

[0014] a feature data acquisition unit, configured to perform feature extraction on the first historical data based on the target feature, to obtain first feature data;

[0015] a first report generation unit, configured to obtain evaluation data based on the first feature data and the target case data, and generate a first evaluation report based on the evaluation data.

[0016] In a third aspect, an embodiment of the present specification provides a computer program product, which stores at least one instruction, and the at least one instruction is suitable for being loaded by a processor and performing the method steps described above.

[0017] In a fourth aspect, an embodiment of the present specification provides a computer storage medium, which stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and performing the method steps described above.

[0018] In a fifth aspect, an embodiment of the present specification provides an electronic device, which comprises a processor and a memory, and the memory stores a computer program, and the computer program is suitable for being loaded by the processor and performing the method steps described above.

[0019] In the embodiments of the present specification, by acquiring target case data of a target case type, performing feature extraction on the target case data by using a feature extraction model to obtain target features, acquiring first historical data based on the target features, performing feature extraction based on the first historical data to obtain first feature data, obtaining evaluation data based on the first feature data and the target case data, and generating a first evaluation report, the historical data is obtained based on the features extracted from the target case data, the evaluation report corresponding to the target case type is generated, and thus the time required for analyzing cases and writing reports is reduced, the processing efficiency of case analysis is improved, and the efficiency and accuracy of case analysis are ensured. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present specification or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present specification, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0021] Figure 1 a system architecture diagram of a case analysis method provided by an embodiment of the present specification;

[0022] Figure 2A flowchart of a case analysis method provided by an embodiment of the present specification is shown in FIG. 1.

[0023] Figure 3 An example schematic diagram of first historical data provided by an embodiment of the present specification is shown in FIG. 2.

[0024] Figure 4 An example schematic diagram of first feature data provided by an embodiment of the present specification is shown in FIG. 3.

[0025] Figure 5 An example schematic diagram of generating a third evaluation report provided by an embodiment of the present specification is shown in FIG. 4.

[0026] Figure 6 A flowchart of a case analysis method provided by an embodiment of the present specification is shown in FIG. 1.

[0027] Figure 7 A structural schematic diagram of a case analysis device provided by an embodiment of the present specification is shown in FIG. 5.

[0028] Figure 8 A structural schematic diagram of a feature extraction unit provided by an embodiment of the present specification is shown in FIG. 6.

[0029] Figure 9 A structural schematic diagram of a first report generation unit provided by an embodiment of the present specification is shown in FIG. 7.

[0030] Figure 10 A structural schematic diagram of a case analysis device provided by an embodiment of the present specification is shown in FIG. 5.

[0031] Figure 11 A structural schematic diagram of an electronic device provided by an embodiment of the present specification is shown in FIG. 8. DETAILED DESCRIPTION

[0032] In order to make the features and advantages of the present specification more obvious and easy to understand, the technical solutions in the embodiments of the present specification will be described clearly and completely below with reference to the drawings in the embodiments of the present specification. Obviously, the described embodiments are only some of the embodiments of the present specification, but not all the embodiments. Based on the embodiments in the present specification, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present specification.

[0033] In the prior art, when analyzing a case, the problems of complex data required to be processed, real-time updating of data, long analysis time, etc. exist in artificial analysis of the case, which leads to low efficiency of case analysis and cannot meet the high efficiency and accuracy required by case analysis.

[0034] Based on this, the case analysis method provided in the embodiments of the present specification is provided. By obtaining target case data of a target case type, the target features are extracted from the target case data by using a feature extraction model, the first historical data is obtained based on the target features, the first feature data is obtained by extracting features based on the first historical data, and the first evaluation report is generated after the evaluation data is obtained based on the first feature data and the target case data. Therefore, the historical data is obtained based on the features extracted from the target case data, and the evaluation report corresponding to the target case type is generated, thereby reducing the time required for analyzing cases and writing reports, improving the processing efficiency of case analysis, and ensuring the efficiency and accuracy of case analysis.

[0035] Please refer to Figure 1 A system structure diagram of case analysis is provided for the embodiments of the present specification. As shown in Figure 1 The case analysis method provided in the embodiments of the present specification can be applied to a terminal device to realize the process of generating an evaluation report for case analysis. The system structure provided in the embodiments of the present specification mainly includes a terminal device 10 and a data center 20. The terminal device 10 can be a device with a feature extraction model and storing case data, used for feature extraction and report generation of cases. Specifically, it can be a desktop computer, a notebook computer, a tablet computer, or other electronic devices with data display and data processing functions. It can also be an independent server or a server cluster composed of multiple servers used by an enterprise, including but not limited to hardware servers, virtual servers, cloud servers, and microcomputers such as personal computers. The data center 20 can be a data storage device storing case-related data. Specifically, it can be an independent server or a server cluster composed of multiple servers used by an enterprise, including but not limited to hardware servers, virtual servers, cloud servers, and microcomputers such as personal computers.

[0036] In the embodiments of the present specification, after the terminal device 10 obtains the target case data, the target features are obtained by extracting features from the target case data using a feature extraction model. The corresponding first historical data is obtained from the data center 20 based on the target features. The first feature data is obtained by refining based on the first historical data and the target features. The evaluation data is obtained based on the first feature data and the target case data, and then the first evaluation report is generated.

[0037] In the embodiment of the present specification, by acquiring target case data of a target case type, a feature extraction model is used to extract features from the target case data to obtain target features, first historical data is acquired based on the target features, first feature data is extracted based on the first historical data, and after the evaluation data is obtained based on the first feature data and the target case data, a first evaluation report is generated, so as to realize the extraction of features based on the target case data to acquire historical data and generate an evaluation report corresponding to the target case type, thereby reducing the time required for analyzing cases and writing reports, improving the processing efficiency of case analysis, and ensuring the efficiency and accuracy of case analysis.

[0038] Based on Figure 1 the system architecture shown in FIG. 1, the case analysis method provided by the embodiment of the present specification will be described in detail. Figures 2-5 The case analysis method provided by the embodiment of the present specification will be described in detail.

[0039] Please refer to Figure 2 a flowchart of a case analysis method provided by the embodiment of the present specification. As shown in FIG. 2, the method can include the following steps S102-S110. Figure 2

[0040] S102, acquiring target case data corresponding to a target case type, and inputting the target case data into a pre-trained feature extraction model;

[0041] In one embodiment, the target case type can be the type of the case that needs to be analyzed, for example, it can be an insurance accident case, a regional camera distribution scheme, etc.

[0042] The target case data can be case data corresponding to the target case type, and the target case data can include the content of at least one case, for example, when the target case type is an insurance accident case, one of the case data in the target case data includes the accident time, accident location, accident reason, accident process, etc. related to the case data. The method of obtaining the target case data can be pre-stored data in the terminal device, data input by the clerk performing case analysis, or data obtained by the terminal device from the case database.

[0043] It should be noted that, in order to ensure the reliability and generality of the analysis result of the target case type, the number of target case data can be set to a minimum number, for example, three hundred cases, etc. Further, in order to improve the practicality of the target case data, the content of each case in the target case data needs to ensure the difference, thereby improving the range of cases covered by the target case data.

[0044] ​Further, the feature extraction model can be a pre-trained model for extracting features of the target case data. The feature extraction model can be a model obtained by adaptively training a text analysis model for the target case type. For example, the target case type is an insurance accident case, and the feature extraction model for analyzing the target case type can be a text analysis model pre-trained according to the target case data related to the insurance accident type. When the feature extraction capability of the text analysis model for the target case data meets the expected requirement, the text analysis model is considered to be trained and can be used as the feature extraction model for extracting features of the target case data related to the insurance accident type. The specific training method can be adjusted according to the actual situation.

[0045] S104, extracting features of the target case data by using the feature extraction model to obtain target features of the target case data;

[0046] In one embodiment, the feature extraction model is used to extract candidate features from the target case data, and the feature extraction model is used to screen the candidate features based on data explainability to obtain target features of the target case data.

[0047] The candidate features can be all words or sentences with feature properties included in the target case data, such as accident time, accident location, whether to report an alarm, etc.

[0048] The target features can be features obtained by screening and analyzing the candidate features based on explainability. The data explainability can be a data analysis method. Based on the data explainability, the candidate features with practical value can be obtained, and the candidate features with practical value are used as the target features. For example, the candidate features include accident time, accident location, accident cause, whether to report an alarm, accident process, and relationship between the reporter and the person involved in the accident. Since the purpose of the target case data analysis is to reduce the accident probability, the accident cause and the accident process are screened and analyzed based on the data explainability, and the main features of the accident are determined to be driving a non-motor vehicle and causing an accident. Therefore, the non-motor vehicle can be used as the target feature.

[0049] It should be noted that the feature screening method based on explainability needs to be adjusted according to the target case type of the target case data to screen the target features required by the target case data. The adjustment of the explainability can be performed when the feature extraction model is trained.

[0050] S106, obtaining first historical data corresponding to the target features from a data center associated with the target case type;

[0051] In an embodiment, after the target feature is obtained, the first historical data corresponding to the target feature is obtained in the data center associated with the target case type based on the target feature.

[0052] The data center can be a database storing data associated with the target case type, and different target case types correspond to different data centers. For example, when the target case type is an insurance accident type, the corresponding data center can be a database of a network shopping platform; when the target case type is a regional camera distribution scheme, the corresponding data center can be a case library of the region, which includes relevant data of traffic accidents, theft events, etc. that have occurred in the region.

[0053] For example, a feasible method for obtaining first historical data from a data center can be that the terminal device sends a data request carrying a target feature to the data center, the data center searches the stored data based on the target feature, filters out the data associated with the target feature, and returns the data as first historical data to the terminal device.

[0054] The first historical data can include data associated with the target feature, and different target case data correspond to different contents included in the first historical data.

[0055] For example, when the target case type is an insurance accident case, and the target feature extracted from the target case data of the target case type includes a non-motor vehicle, the first historical data is obtained from the database of a transaction platform capable of trading non-motor vehicles and related accessories of non-motor vehicles. The obtained first historical data can include transaction records of different users, which can include transaction time, transaction amount, item name, etc. For example, as shown in Figure 3 Figure 3 which includes the sunshades and helmets purchased by users A, B and C on January 1, 2023, and the corresponding transaction amounts.

[0056] It can be understood that since ordinary users rarely purchase non-motor vehicles, the frequency of purchasing related accessories after purchasing non-motor vehicles will increase after purchasing non-motor vehicles, so purchasing non-motor vehicles can be used as auxiliary evidence of purchasing or using non-motor vehicles.

[0057] ​It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.), and signals involved in the embodiments of this specification are all authorized by the user or fully authorized by all parties, and the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. For example, transaction information involved in this specification was obtained under full authorization.

[0058] S108, Based on the target features, the first historical data is extracted to obtain the first feature data;

[0059] In one embodiment, the first feature data can be data that can be used for case analysis, obtained by extracting features from first historical data based on the target feature. For example, when the target feature is a non-motorized vehicle, the first feature data can be the number of transactions and the total transaction amount, etc.

[0060] A feasible method for feature extraction from first historical data is to filter the first historical data based on a preset time period to obtain target historical data in the first historical data, and then extract features from the target historical data based on the target features to obtain the first feature data corresponding to the target features.

[0061] For example, the preset duration is 7 days, the target feature is non-motorized vehicles, after obtaining the first historical data, the first historical data is filtered to obtain the historical data of the next 7 days at the current time as the target historical data, and then the target historical data is refined based on the feature of non-motorized vehicles to obtain the first feature data corresponding to the target feature.

[0062] For example, such as Figure 4 As shown, Figure 4 To Figure 3 The table obtained after feature extraction. It can be seen that, since the target feature is non-motorized vehicles, therefore... Figure 3 When performing feature extraction, the number of transactions and the total transaction amount within a preset time period are obtained based on the target features.

[0063] It should be noted that, Figure 3 and Figure 4 For example, when the target case type is an insurance claim, if the target case type is a regional camera distribution plan, feature extraction based on target characteristics could involve extracting the number of cases occurring in each region within a preset time period from the target's historical data. The specific extraction method can be set according to the actual situation of the target characteristics.

[0064] The preset time length can be a time length set according to the target case type, for example, can be 7 days, 30 days or one year, etc. The specific time length of the preset time length can be set according to actual needs.

[0065] It can be understood that, in order to improve the practicability of the first feature data, the obtained first historical data is filtered based on the preset time length to obtain target historical data, so as to avoid negative effects caused by analyzing the target case data due to the too long time node spanned by the first historical data. For example, the data center stores a transaction record of user D for a non-motor vehicle, but the transaction record is three years ago, so it is considered that the transaction record has no effect on the current user D, and it is not necessary to obtain this transaction record to increase the calculation amount of data analysis.

[0066] S110, based on the first feature data and the target case data, obtaining evaluation data, and generating a first evaluation report based on the evaluation data;

[0067] In one embodiment, after obtaining the first feature data, the first feature data is classified based on the classification rule corresponding to the target case type to obtain the feature type included in the first feature data, the case occurrence probability corresponding to each feature type in the target case data is obtained, the evaluation data corresponding to the target case data is determined based on the case occurrence probability, and the first evaluation report corresponding to the target case type is generated based on the evaluation data.

[0068] The classification rule can be a rule determined based on the target case type, and the classification rules of different target case types can be the same or different. One feasible method can be to classify according to the number of transactions within the preset time length, and set different thresholds to classify the number of transactions. For example, ten times are taken as the threshold for classification, and ten categories are divided, which are zero to ten times, eleven to twenty times, …, up to ninety times or more, a total of ten categories. The specific classification method can be set according to data needs.

[0069] The feature type can be the type of each category obtained after classification, for example, zero to ten times is one of the types of the first feature data.

[0070] The case occurrence probability can be used to indicate the possibility of occurrence of a case for each feature type, for example, in the feature type of zero to ten times, the proportion of users who have occurred a case among the objects in the first feature data of this feature type is obtained. For example, there are three hundred users in the feature type of zero to ten times, and fifteen of them have occurred a case, so the case occurrence probability of this feature type is five percent.

[0071] The evaluation data can be data obtained by evaluating the objects included in the target case data based on the occurrence probability of each feature type. One possible method of obtaining the evaluation data can be as follows: after obtaining the occurrence probability of each feature type, the target objects included in the target case type are classified based on the occurrence probability, the object type corresponding to each target object is determined, and the evaluation data corresponding to the target case data is obtained based on the object type.

[0072] The target object can be a user or a region determined according to the target case type, for example, if the target case type is an insurance risk type, the target object corresponding to the target case data of the target case type can be a user who has a risk case; if the target case type is a regional camera distribution scheme, the target object can be each region.

[0073] The object type can be a type determined by classifying the target object based on the occurrence probability. For example, if the target case type is an insurance risk type and the feature type is the number of transactions, which is eleven to twenty times, the occurrence probability is forty-five percent, then according to the occurrence probability, the target object included in the feature type can be determined as a user prone to risk.

[0074] Further, after determining the object type, the evaluation results of each feature type in the target case data are obtained based on the object type, and the evaluation results are used as the evaluation data corresponding to the target case data.

[0075] Further, after obtaining the evaluation data, the target case data is arranged based on the evaluation data to generate a first evaluation report corresponding to the target case type.

[0076] For example, the evaluation data in the evaluation data includes the evaluation results corresponding to each feature type in the target case data, and the evaluation data of the target case data is converted into an evaluation article by a pre-trained text generation model, and the evaluation article is used as the first evaluation report of the target case type. For example, the evaluation data of the target case data can be generated as SQL (Structured Query Language), or as a paper report, etc.

[0077] The specific form of the first evaluation report can be set according to actual needs, which can be SQL, a paper, etc.

[0078] Further, in order to avoid insufficient first historical data from being obtained from the data center according to the target feature, thereby improving the accuracy and reliability of the evaluation of the target case type, a feasible method can be that case inquiry data is obtained, case features related to the target case data are output based on the case inquiry data, second historical data is obtained, a second evaluation report is generated based on the second historical data and the case features, a third evaluation report is generated based on the first evaluation report and the second evaluation report, and the third evaluation report is taken as the final output evaluation report corresponding to the target case type.

[0079] The case inquiry data can be data input by a clerk who analyzes the case to the database. The case inquiry data can be a question sentence with inquiry meaning. The database includes pre-input target case data. After the case inquiry data is obtained, the terminal device obtains case features related to the case inquiry data from the database through a feature extraction model in response to the case inquiry data.

[0080] The second historical data can be historical data obtained by the clerk based on the case features by querying the data center. The method for the clerk to obtain the second historical data can be to manually query the case features through a query portal of the data center. It can be understood that the second historical data is data obtained by the clerk manually, and the specific obtaining method can be selected according to the actual situation of the clerk.

[0081] The second evaluation report can be an evaluation report generated based on the second historical data and the case features. Specifically, the second historical data can be obtained, second feature data can be obtained based on the case features and the second historical data, and a second evaluation report can be generated based on the second feature data and the target case data. The generation method of the second evaluation report can refer to the generation method of the first evaluation report, which will not be described here.

[0082] The third evaluation report can be a report generated by integrating the first evaluation report and the second evaluation report. Specifically, the same content in the first evaluation report and the second evaluation report can be copied into a blank report to generate the third evaluation report. For the different parts in the first evaluation report and the second evaluation report, the updated content in the second evaluation report is selected and added to the third evaluation report, thereby generating the third evaluation report corresponding to the target case type.

[0083] It should be noted that when there is a logical conflict in the first evaluation report and the second evaluation report, the part generated by analyzing the content is added to the third evaluation report.

[0084] Further, the third evaluation report generation method can be used to obtain the first historical data, the target feature, the second historical data and the case feature, obtain the second evaluation data based on the first historical data, the target feature, the second historical data and the case feature, and generate the third evaluation report based on the second evaluation data. The third evaluation report generation method can refer to the first evaluation report generation method, which will not be described here. The specific generation method can be set according to actual needs.

[0085] As shown in Figure 5 , Figure 5 The third evaluation report generation method is shown in method 1, which is based on the first evaluation report and the second evaluation report to generate the third evaluation report; method 2 is based on the first historical data, the target feature, the second historical data and the case feature to generate the third evaluation report.

[0086] Further, based on the first evaluation report, the target case type related case can be prevented. For example, if the target case type is an insurance risk type, when the first evaluation report indicates that the object type corresponding to the feature type A has a high risk probability, the object of the object type needs to be transacted in the insurance transaction, the transaction amount is increased or the transaction is refused, etc.

[0087] In the embodiments of the present specification, by obtaining the target case data of the target case type, the feature extraction model is used to extract the target feature from the target case data, the first historical data is obtained based on the target feature, the first feature data is obtained based on the first example data, and the evaluation data is obtained based on the first feature data and the target case data. After generating the first evaluation report, the features extracted based on the target case data are obtained to obtain the historical data, generate the evaluation report corresponding to the target case type, and further reduce the time required for analyzing the case and writing the report, improve the processing efficiency of the case analysis, and ensure the efficiency and accuracy of the case analysis.

[0088] Please refer to Figure 6 , a flowchart of a case analysis method is provided in the embodiments of the present specification. As shown in Figure 6 , the method can include the following steps S202-S222.

[0089] S202, obtaining target case data corresponding to a target case type, and inputting the target case data into a pre-trained feature extraction model;

[0090] In one embodiment, the target case type can be the type of case that needs to be analyzed, for example, it can be an insurance risk case, a regional camera distribution scheme, etc.

[0091] The target case data can be case data corresponding to the target case type. The target case data can include case content of at least one case. For example, when the target case type is an insurance accident case, one of the case data in the target case data includes accident time, accident location, accident reason, accident process, etc.

[0092] It should be noted that, in order to ensure the reliability and universality of the analysis result of the target case type, the number of target case data can be set to a minimum number, for example, three hundred cases, etc. Further, in order to improve the practicability of the target case data, the content of each case in the target case data needs to ensure the difference, thereby improving the case range covered by the target case data.

[0093] Further, the feature extraction model can be a pre-trained model for extracting features of the target case data. The feature extraction model can be a model obtained by adaptively training a text analysis model for the target case type. For example, the target case type is an insurance accident case, and the feature extraction model for analyzing the target case type can be a text analysis model pre-trained according to the target case data related to the insurance accident type. When the feature extraction capability of the text analysis model for the target case data meets the expected requirement, the text analysis model is considered to be trained and can be used as a feature extraction model for the target case data of the insurance accident type. The specific training method can be adjusted according to the actual situation.

[0094] S204, using the feature extraction model to extract features of the target case data to obtain candidate features in the target case data;

[0095] In one embodiment, the candidate features can be all words or sentences with characteristic properties included in the target case data, such as accident time, accident location, whether to call the police, etc.

[0096] For example, the target case data includes: accident time: 20230101; accident location: A province B city urban area C street D community gate entrance pedestrian crossing; accident reason: XXX was hit by a riding motorbike by a minor, has reported to the police, failed to rescue in the hospital, died; death time: 20230101-20:34; accident process: 20230101-20:34; whether to call 120: yes; whether to call the police: yes, etc. The candidate features obtained by using the feature extraction model for feature extraction can be accident time, accident location, accident reason, accident process, whether to call 120, and whether to call the police.

[0097] S206, performing feature screening on the candidate features based on data explainability by using the feature extraction model to obtain target features of the target case data;

[0098] In one embodiment, the target features can be features obtained by screening and analyzing the candidate features based on explainability.

[0099] The data explainability can be a data analysis method, and screening the candidate features based on the data explainability can obtain features with practical value in the candidate features, and the features with practical value are taken as the target features. For example, the candidate features include the time of the accident, the location of the accident, the cause of the accident, whether the alarm is reported, the course of the accident, the relationship between the reporter and the person involved in the accident, etc. Since the purpose of the target case data analysis is to reduce the probability of the accident, the cause of the accident, the course of the accident, etc. and the features that affect the probability of the accident can be screened out for analysis according to the data explainability, and the main features of the accident are determined as driving a non-motor vehicle to cause a traffic accident, and then the non-motor vehicle can be taken as the target feature.

[0100] It should be noted that the method of explainability for feature screening needs to be adjusted according to the target case type of the target case data to screen out the target features required by the target case data. The adjustment of explainability can be performed when the feature extraction model is trained.

[0101] S208, obtaining first historical data corresponding to the target features in the data center associated with the target case type;

[0102] In one embodiment, after obtaining the target features, the first historical data corresponding to the target features is obtained in the data center associated with the target case type based on the target features.

[0103] The data center can be a database storing data associated with the target case type, and different target case types correspond to different data centers. For example, when the target case type is an insurance accident type, the corresponding data center can be a database of a network shopping platform; when the target case type is a regional camera distribution scheme, the corresponding data center can be a case library of the region, which includes relevant data of traffic accidents, theft events, etc. that have occurred in the region.

[0104] For example, a feasible method for obtaining the first historical data from the data center can be that the terminal device sends a data request carrying the target features to the data center, the data center searches the stored data based on the target features, screens out the data associated with the target features, and returns the data as the first historical data to the terminal device.

[0105] The first historical data can include data associated with the target feature. The first historical data corresponding to different target case data includes different contents.

[0106] For example, the target case type is an insurance accident case, and the target feature extracted from the target case data according to the target case type includes a non-motor vehicle. The first historical data is obtained from the database of a transaction platform capable of trading non-motor vehicles and related accessories of non-motor vehicles. The obtained first historical data can include transaction records of different users, which can include transaction time, transaction amount, item name, etc. For example, it can include Figure 3 Figure 3 user A, B and C purchased sunshades and helmets on January 1, 2023, respectively, and the corresponding transaction amounts.

[0107] It can be understood that the frequency of purchasing a non-motor vehicle by an ordinary user is very low, but the frequency of purchasing related accessories after purchasing a non-motor vehicle will increase after purchasing a non-motor vehicle. Therefore, purchasing a non-motor vehicle can be used as an auxiliary basis for purchasing or using a non-motor vehicle.

[0108] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data for analysis, stored data, displayed data, etc.) and signals involved in the embodiments of the present specification are all authorized by the user or fully authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions. For example, the transaction information and the like involved in the present specification are obtained under full authorization.

[0109] In S210, the first historical data is filtered based on a preset time length to obtain target historical data in the first historical data.

[0110] In one embodiment, the target historical data can be data obtained by filtering the first historical data based on a preset time length. The specific content of the target historical data can be determined based on the specific time length of the preset time length.

[0111] For example, the preset time length is 7 days, the target feature is a non-motor vehicle, and after obtaining the first historical data, the first historical data is filtered to obtain historical data of the adjacent 7 days of the current time as target historical data corresponding to the target feature.

[0112] The preset time length can be a time length set according to the target case type, for example, 7 days, 30 days or one year, etc. The specific time length of the preset time length can be set according to actual needs.

[0113] ​Understandably, to improve the usability of the primary feature data, the acquired primary historical data is filtered based on a preset time frame to obtain target historical data. This avoids negative impacts on the analysis of target case data due to excessively long time spans in the primary historical data. For example, if the data center stores a transaction record of user D for a non-motorized vehicle, but the transaction record is from three years ago, it is considered that the transaction record has no impact on user D at the current moment, and there is no need to obtain this transaction record, thus avoiding increasing the computational load of data analysis.

[0114] S212, Based on the target features, feature extraction is performed on the target historical data to obtain the first feature data corresponding to the target features;

[0115] In one embodiment, the first feature data can be data that can be used for case analysis, obtained by extracting features from first historical data based on the target feature. For example, when the target feature is a non-motorized vehicle, the first feature data can be the number of transactions and the total transaction amount, etc.

[0116] Based on the target features, feature extraction is performed on the target's historical data to obtain the first feature data corresponding to the target features.

[0117] For example, the preset duration is 7 days, the target feature is non-motorized vehicles, after obtaining the first historical data, the first historical data is filtered to obtain the historical data of the next 7 days at the current time as the target historical data, and then the target historical data is refined based on the feature of non-motorized vehicles to obtain the first feature data corresponding to the target feature.

[0118] For example, such as Figure 4 As shown, Figure 4 To Figure 3 The table obtained after feature extraction. It can be seen that, since the target feature is non-motorized vehicles, therefore... Figure 3 When performing feature extraction, the number of transactions and the total transaction amount within a preset time period are obtained based on the target features.

[0119] It should be noted that, Figure 3 and Figure 4 For example, when the target case type is an insurance claim, if the target case type is a regional camera distribution plan, feature extraction based on target characteristics could involve extracting the number of cases occurring in each region within a preset time period from the target's historical data. The specific extraction method can be set according to the actual situation of the target characteristics.

[0120] S214, classify the first feature data based on the classification rules corresponding to the target case type to obtain the feature types included in the first feature data;

[0121] In an embodiment, the classification rule can be a rule determined based on a target case type, and the classification rules of different target case types can be the same or different.

[0122] For example, a feasible classification method can be to classify according to the number of transactions in a preset time period, and set different thresholds to classify the number of transactions. For example, ten times is set as the threshold for classification, and is divided into ten categories, i.e., zero to ten times, eleven to twenty times,..., and ninety or more times, a total of ten categories. The specific classification method can be set according to data needs.

[0123] The feature type can be the type of each category obtained after classification, for example, zero to ten times is one of the types of the first feature data.

[0124] S216, obtaining a case occurrence probability corresponding to each feature type in the target case data;

[0125] In an embodiment, the case occurrence probability can be used to indicate the possibility of occurrence of a case for each feature type, for example, in the feature type of zero to ten times, the proportion of users in the first feature data that have occurred a case is obtained.

[0126] For example, there are a total of three hundred users in the feature type of zero to ten times, and fifteen of them have occurred a case, so the case occurrence probability of this feature type is five percent.

[0127] S218, classifying target objects included in the target case type based on the case occurrence probability to determine an object type corresponding to each target object;

[0128] In an embodiment, the target object can be a user or a region determined according to the target case type, for example, the target case type is an insurance risk type, and the target object corresponding to the target case data of the target case type can be a user who has occurred a risk case; if the target case type is a region camera distribution scheme, the target object can be each region.

[0129] The object type can be a type determined after classifying the target object based on the case occurrence probability. For example, the target case type is an insurance risk type, the feature type is the number of transactions of eleven to twenty times, and the case occurrence probability is forty-five percent, so the target object included in the log type can be determined as a user prone to risk according to the case occurrence probability.

[0130] S220, obtaining evaluation data corresponding to the target case data based on the object type;

[0131] In an embodiment, after the object type is determined, the evaluation data can be obtained based on the object type to obtain the evaluation result of each feature type in the target case data, and the evaluation result is taken as the evaluation data corresponding to the target case data.

[0132] For example, when the target case type is an insurance accident type, the case probability corresponding to each feature type included in the target case data is obtained, the object type of the target object in the target case data is determined based on the case probability, and the object type corresponding to each feature type is sorted to obtain the evaluation data corresponding to the target case data.

[0133] S222, generating a first evaluation report corresponding to the target case type based on the evaluation data;

[0134] In an embodiment, the first evaluation report can be a report for evaluating the case occurrence probability of the target case type, and the specific form of the first evaluation report can be set according to actual needs, which can be SQL or a paper.

[0135] For example, the evaluation data of the target case data is converted into an evaluation article by a pre-trained text generation model, and the evaluation article is taken as the first evaluation report of the target case type. For example, the evaluation data of the target case data is generated as SQL (Structured Query Language), or a paper report.

[0136] Further, in order to avoid that when the first historical data is obtained from the data center according to the target feature, the first historical data is insufficient to comprehensively evaluate the target case data, and to improve the accuracy and reliability of the evaluation of the target case type, a feasible method can be that the case inquiry data is obtained, the case features related to the target case data are output based on the case inquiry data, the second historical data is obtained, the second evaluation report is generated based on the second historical data and the case features, the third evaluation report is generated based on the first evaluation report and the second evaluation report, and the third evaluation report is taken as the final output evaluation report corresponding to the target case type.

[0137] The case inquiry data can be the data input by the transaction officer who analyzes the case, and the case inquiry data can be a question sentence with inquiry meaning. The database includes pre-input target case data. After the case inquiry data is obtained, the terminal device obtains the case features related to the case inquiry data from the database in response to the case inquiry data.

[0138] The second historical data can be historical data obtained by the case officer based on the case characteristics after querying in the data center. The case officer can obtain the second historical data by manually querying the case characteristics through the query portal of the data center. It can be understood that the second historical data is obtained by manual query by the case officer, and the specific obtaining method can be selected according to the actual situation of the case officer.

[0139] The second evaluation report can be an evaluation report generated based on the second historical data and the case characteristics. Specifically, the second evaluation report can be generated by obtaining the input second historical data, obtaining second feature data based on the case characteristics and the second historical data, and generating the second evaluation report based on the second feature data and the target case data. The generation method of the second evaluation report can refer to the generation method of the first evaluation report, which will not be described here.

[0140] The third evaluation report can be a report generated by integrating the first evaluation report and the second evaluation report. Specifically, the third evaluation report can be generated by copying the same content in the first evaluation report and the second evaluation report into a blank report, selecting the updated content in the second evaluation report and adding it to the third evaluation report, thereby generating the third evaluation report corresponding to the target case type.

[0141] It should be noted that when there is a logical conflict in the first evaluation report and the second evaluation report, the part generated by analyzing the content is added to the third evaluation report.

[0142] Further, the generation method of the third evaluation report can be to obtain the first historical data, the target characteristics, the second historical data and the case characteristics, obtain the second evaluation data based on the first historical data, the target characteristics, the second historical data and the case characteristics, and generate the third evaluation report based on the second evaluation data. The generation method of the third evaluation report can refer to the generation method of the first evaluation report, which will not be described here. The specific generation method can be set according to actual needs.

[0143] For example, as shown in Figure 5 , the generation method of the third evaluation report is shown in Figure 5 . Method 1 is to generate the third evaluation report by integrating the first evaluation report and the second evaluation report; method 2 is to generate the third evaluation report based on the first historical data, the target characteristics, the second historical data and the case characteristics.

[0144] Furthermore, after generating the first assessment report, preventative actions can be taken for cases related to the target case type based on the report. For example, if the target case type is an insurance claim type, and the first assessment report indicates that the object type corresponding to feature type A has a high probability of being insured, then when an object of that object type wants to conduct an insurance transaction, the transaction amount can be increased or the transaction can be refused.

[0145] In the embodiments of this specification, target case data of the target case type is obtained, and a feature extraction model is used to extract features from the target case data to obtain target features. First historical data is obtained based on the target features, and first feature data is obtained based on the first case data. After obtaining evaluation data based on the first feature data and the target case data, a first evaluation report is generated. A second evaluation report is then obtained through inquiry and data search. A third evaluation report is generated based on the first and second evaluation reports. This achieves the acquisition of historical data based on features extracted from the target case data, the generation of a first evaluation report corresponding to the target case type, and the generation of a second evaluation report based on the second evaluation report obtained after inquiry. This reduces the time required for case analysis and report writing, improves the processing efficiency of case analysis, and ensures the efficiency and accuracy of case analysis by synthesizing multiple evaluation reports.

[0146] based on Figure 1 The system architecture shown below will be combined with... Figures 7-10 This specification provides a detailed description of the case analysis device provided in the embodiments. It should be noted that... Figures 7-10 The case analysis device described herein is used to execute the instructions. Figures 2-6 The methods shown in the embodiments are illustrated for ease of explanation, showing only the parts related to the embodiments of this specification. For specific technical details not disclosed, please refer to this specification. Figures 2-6 The example shown.

[0147] Please see Figure 7 This is a schematic diagram of the structure of a case analysis device provided in the embodiments of this specification. Figure 7 As shown, the case analysis device 1 described in the embodiments of this specification may include: a case data acquisition unit 11, a feature extraction unit 12, a historical data acquisition unit 13, a feature data acquisition unit 14, and a report generation unit 15.

[0148] The case data acquisition unit 11 is used to acquire target case data corresponding to the target case type and input the target case data into a pre-trained feature extraction model.

[0149] Feature extraction unit 12 is used to extract features from the target case data using the feature extraction model to obtain target features of the target case data;

[0150] The historical data acquisition unit 13 is configured to acquire first historical data corresponding to the target feature from a data center associated with the target case type;

[0151] The feature data acquisition unit 14 is configured to perform feature extraction on the first historical data based on the target feature, to obtain first feature data;

[0152] The first report generation unit 15 is configured to obtain evaluation data based on the first feature data and the target case data, and generate a first evaluation report based on the evaluation data.

[0153] Optionally, as shown in Figure 8 The feature extraction unit 12 includes:

[0154] The candidate feature extraction sub-unit 121 is configured to perform feature extraction on the target case data by using the feature extraction model, to obtain candidate features in the target case data;

[0155] The target feature acquisition sub-unit 122 is configured to perform feature screening on the candidate features based on data interpretability by using the feature extraction model, to obtain target features of the target case data.

[0156] Optionally, the feature data acquisition unit 14 is further configured to:

[0157] Filter the first historical data based on a preset time length, to obtain target historical data in the first historical data;

[0158] Perform feature extraction on the target historical data based on the target feature, to obtain first feature data corresponding to the target feature.

[0159] Optionally, as shown in Figure 9 The first report generation unit 15 includes:

[0160] The classification sub-unit 151 is configured to classify the first feature data based on a classification rule corresponding to the target case type, to obtain feature types included in the first feature data;

[0161] The evaluation sub-unit 152 is configured to acquire an occurrence probability corresponding to each of the feature types in the target case data, and determine evaluation data corresponding to the target case data based on the occurrence probability;

[0162] The report generation sub-unit 153 is configured to generate a first evaluation report corresponding to the target case type based on the evaluation data.

[0163] Optionally, the evaluation sub-unit 152 is further configured to:

[0164] obtain an occurrence probability corresponding to each feature type in the target case data;

[0165] classify target objects included in the target case type based on the occurrence probability, and determine an object type corresponding to each target object;

[0166] obtain evaluation data corresponding to the target case data based on the object type.

[0167] Optionally, as shown in Figure 10 The case analysis device 1 further includes:

[0168] The case feature acquisition unit 16 is configured to obtain case inquiry data, and output case features related to the target case data based on the case inquiry data.

[0169] The second report generation unit 17 is configured to obtain input second historical data, and generate a second evaluation report based on the second historical data and the case features.

[0170] The third report generation unit 18 is configured to generate a third evaluation report based on the first evaluation report and the second evaluation report.

[0171] Optionally, the second report generation unit 17 is further configured to:

[0172] obtain input second historical data, and obtain second feature data based on the case features and the second historical data;

[0173] generate a second evaluation report based on the second feature data and the target case data.

[0174] In the embodiments of the present disclosure, by obtaining target case data of a target case type, a feature extraction model is used to extract features from the target case data to obtain target features, first historical data is obtained based on the target features, first feature data is obtained by performing feature extraction based on the first historical data, and a first evaluation report is generated based on the first feature data and the target case data. Further, a second evaluation report is obtained through inquiry and data searching, a third evaluation report is generated based on the first evaluation report and the second evaluation report, thereby realizing feature extraction based on target case data to obtain historical data, generating a first evaluation report corresponding to the target case type, and generating a second evaluation report based on the second evaluation report obtained after inquiry, thereby reducing the time required for analyzing cases and writing reports, improving the processing efficiency of case analysis, and ensuring the efficiency and accuracy of case analysis through the synthesis of multiple evaluation reports.

[0175] The embodiment of the present specification also provides a computer storage medium, which can store a plurality of program instructions, the program instructions being suitable for being loaded and executed by a processor to perform the method steps of the above-described Figures 1-6 The specific implementation process of the method steps of the above-described Figures 1-6 The specific implementation process of the method steps of the above-described

[0176] The embodiment of the present specification also provides a computer program product, which stores at least one instruction, the at least one instruction being loaded and executed by the processor to perform the case analysis method of the above-described Figures 1-6 The specific implementation process of the method steps of the above-described Figures 1-6 The specific implementation process of the method steps of the above-described

[0177] Referring to Figure 11 The embodiment of the present specification provides a structural schematic diagram of an electronic device. As shown in Figure 11 The electronic device 1000 can include at least one processor 1001, such as a CPU, at least one network interface 1004, an input / output interface 1003, a memory 1005, and at least one communication bus 1002. The communication bus 1002 is used to realize the connection and communication between the components. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a WI-FI interface). The memory 1005 can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The memory 1005 can also be at least one storage device located away from the aforementioned processor 1001. As shown in Figure 11 The memory 1005, as a computer storage medium, can include an operating system, a network communication module, an input / output interface module, and a case analysis application program.

[0178] In the electronic device 1000 shown in Figure 11 The input / output interface 1003 is mainly used to provide an input interface for the user and obtain data input by the user.

[0179] In one embodiment, the processor 1001 can be used to invoke the case analysis application program stored in the memory 1005 and specifically perform the following operations:

[0180] Obtain target case data corresponding to a target case type, and input the target case data into a pre-trained feature extraction model;

[0181] The feature extraction model is used to extract features of the target case data, to obtain target features of the target case data;

[0182] acquire first historical data corresponding to the target feature in a data center associated with the target case type;

[0183] perform feature extraction on the first historical data based on the target feature, to obtain first feature data;

[0184] obtain evaluation data based on the first feature data and the target case data, and generate a first evaluation report based on the evaluation data.

[0185] Optionally, when the processor 1001 performs feature extraction on the target case data by using the feature extraction model to obtain target features of the target case data, the following operations are specifically performed:

[0186] perform feature extraction on the target case data by using the feature extraction model, to obtain candidate features in the target case data;

[0187] perform feature screening on the candidate features based on data explainability by using the feature extraction model, to obtain target features of the target case data.

[0188] Optionally, when the processor 1001 performs feature extraction on the first historical data based on the target feature to obtain first feature data, the following operations are specifically performed:

[0189] perform screening on the first historical data based on a preset time length, to obtain target historical data in the first historical data;

[0190] perform feature extraction on the target historical data based on the target feature, to obtain first feature data corresponding to the target feature.

[0191] Optionally, when the processor 1001 obtains evaluation data based on the first feature data and the target case data, and generates a first evaluation report based on the evaluation data, the following operations are specifically performed:

[0192] classify the first feature data based on a classification rule corresponding to the target case type, to obtain feature types included in the first feature data;

[0193] obtain occurrence probabilities of each of the feature types in the target case data, and determine evaluation data corresponding to the target case data based on the occurrence probabilities;

[0194] generate a first evaluation report corresponding to the target case type based on the evaluation data.

[0195] Optionally, the processor 1001, in the execution of acquiring the case occurrence probability corresponding to each feature type in the target case data, and determining the evaluation data corresponding to the target case data based on the case occurrence probability, specifically performs the following operations:

[0196] acquiring the case occurrence probability corresponding to each feature type in the target case data;

[0197] classifying the target objects included in the target case type based on the case occurrence probability, and determining the object type corresponding to each target object;

[0198] obtaining the evaluation data corresponding to the target case data based on the object type.

[0199] Optionally, the processor 1001 further performs the following operations:

[0200] acquiring case inquiry data, and outputting case features related to the target case data based on the case inquiry data;

[0201] acquiring input second historical data, and generating a second evaluation report based on the second historical data and the case features;

[0202] generating a third evaluation report based on the first evaluation report and the second evaluation report.

[0203] Optionally, the processor 1001, in the execution of acquiring input second historical data, and generating a second evaluation report based on the second historical data and the case features, specifically performs the following operations:

[0204] acquiring input second historical data, and obtaining second feature data based on the case features and the second historical data;

[0205] generating a second evaluation report based on the second feature data and the target case data.

[0206] In the embodiments of the present application, by obtaining target case data of a target case type, a feature extraction model is used to extract features from the target case data to obtain target features, first historical data is obtained based on the target features, first feature data is obtained by feature extraction based on the first historical data, and after the evaluation data is obtained based on the first feature data and the target case data, a first evaluation report is generated. Further, the second evaluation report is obtained by inquiry and data searching, the third evaluation report is generated based on the first evaluation report and the second evaluation report, thereby realizing the extraction of features based on the target case data to obtain historical data, generating the first evaluation report corresponding to the target case type, and generating the second evaluation report based on the inquiry to obtain the second evaluation report, thereby reducing the time required for analyzing cases and writing reports, improving the processing efficiency of case analysis, and ensuring the efficiency and accuracy of case analysis through the synthesis of multiple evaluation reports.

[0207] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM), a random access memory (RAM), or the like.

[0208] The above only discloses the preferred embodiments of the present application, and of course cannot limit the scope of the rights of the present application. Therefore, equivalent changes made according to the claims of the present application still fall within the scope of the present application.

Claims

1. A case analysis method, the method comprising: obtaining target case data corresponding to a target case type, and inputting the target case data into a pre-trained feature extraction model; extracting features of the target case data by using the feature extraction model to obtain target features of the target case data; obtaining first historical data corresponding to the target features in a data center associated with the target case type; extracting features of the first historical data based on the target features to obtain first feature data; obtaining evaluation data based on the first feature data and the target case data, and generating a first evaluation report based on the evaluation data.

2. The method of claim 1, wherein the extracting features of the target case data by using the feature extraction model to obtain target features of the target case data comprises: extracting candidate features in the target case data by using the feature extraction model; screening the candidate features based on data interpretability by using the feature extraction model to obtain target features of the target case data.

3. The method of claim 1, wherein the extracting features of the first historical data based on the target features to obtain first feature data comprises: screening the first historical data based on a preset time length to obtain target historical data in the first historical data; extracting features of the target historical data based on the target features to obtain first feature data corresponding to the target features.

4. The method of claim 1, wherein the obtaining evaluation data based on the first feature data and the target case data, and generating a first evaluation report based on the evaluation data comprises: classifying the first feature data based on classification rules corresponding to the target case type to obtain feature types included in the first feature data; obtaining occurrence probabilities of each of the feature types in the target case data, and determining evaluation data corresponding to the target case data based on the occurrence probabilities; generating a first evaluation report corresponding to the target case type based on the evaluation data.

5. The method of claim 4, wherein the obtaining occurrence probabilities of each of the feature types in the target case data, and determining evaluation data corresponding to the target case data based on the occurrence probabilities comprises: obtaining occurrence probabilities of each of the feature types in the target case data; classifying target objects included in the target case type based on the occurrence probabilities to determine object types corresponding to each of the target objects; obtaining evaluation data corresponding to the target case data based on the object types.

6. The method of claim 1, further comprising: obtaining case inquiry data, and outputting case features related to the target case data based on the case inquiry data; obtaining input second historical data, and generating a second evaluation report based on the second historical data and the case features; generating a third evaluation report based on the first evaluation report and the second evaluation report.

7. The method of claim 6, wherein the obtaining input second historical data, generating a second evaluation report based on the second historical data and the case characteristics comprises: obtaining input second historical data, obtaining second feature data based on the case characteristics and the second historical data; generating a second evaluation report based on the second feature data and the target case data.

8. A case analysis device, the device comprising: a case data obtaining unit configured to obtain target case data corresponding to a target case type, and input the target case data into a pre-trained feature extraction model; a feature extraction unit configured to extract features of the target case data by using the feature extraction model, and obtain target features of the target case data; a historical data obtaining unit configured to obtain first historical data corresponding to the target features in a data center associated with the target case type; a feature data obtaining unit configured to extract features of the first historical data based on the target features, and obtain first feature data; a first report generating unit configured to obtain evaluation data based on the first feature data and the target case data, and generate a first evaluation report based on the evaluation data.

9. A computer storage medium storing a plurality of instructions, the instructions being adapted to be loaded and executed by a processor to implement the steps of the method of any one of claims 1-7.

10. An electronic device comprising: a processor and a memory; wherein the memory stores a computer program, the computer program being adapted to be loaded and executed by the processor to implement the steps of the method of any one of claims 1-7.

11. A computer program product storing at least one instruction, the at least one instruction being executed by a processor to implement the steps of the method of any one of claims 1-7.

Images