Claim settlement abnormity identification method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, can be applied to business system platforms of medical health, financial science and technology and the like, and discloses a claim settlement abnormity identification method and device, computer equipment and a storage medium. Performing feature extraction on the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information; performing content integration processing on the case information, the contradiction point detection result and a historical claim settlement record of the target customer to obtain associated content of the claim settlement case, and generating an input text of a pre-trained claim settlement anomaly recognition model according to the associated content; based on the input text, generating an anomaly recognition result of the claim settlement case through the claim settlement anomaly recognition model; therefore, the accuracy and efficiency of claim settlement abnormity identification can be effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a claim settlement exception identification method and device, computer equipment and a computer readable storage medium. BACKGROUND

[0002] Currently, in the insurance industry, claim settlement is one of the core links of insurance services. The claim settlement process involves the processing and auditing of a large amount of case information, but the traditional claim settlement processing method mainly relies on manual auditing. This method is not only inefficient, but also susceptible to human factors, leading to inconsistencies and errors in the audit results.

[0003] Currently, with the rapid development of insurance business, the number of claim settlement cases is increasing, and the complexity is also increasing. In this case, the traditional claim settlement processing method faces many challenges:

[0004] 1. Large amount of data: The number of claim settlement cases is large, and manual auditing is difficult to process a large amount of data in a short time, resulting in low efficiency of manual auditing;

[0005] 2. Data complexity: Claim settlement cases involve multiple types of information, including text, numerical and time series data, which require professional technical means for processing and analysis. Manual auditing may lead to a decrease in accuracy;

[0006] 3. Difficulty in identifying abnormalities: Claim settlement abnormalities (such as fraud, incorrect information, etc.) are often hidden in a large amount of normal data, making it extremely difficult for humans to identify abnormalities and easily miss them.

[0007] 4. Low efficiency: Manual auditing is not only time-consuming, but also prone to errors due to fatigue and other factors.

[0008] In the field of medical health, the complexity of insurance claim settlement is particularly prominent. Medical claim settlement cases not only involve a large number of medical records, diagnosis reports and treatment cost details, but also may involve complex medical terminology and professional judgment. These characteristics make it more difficult for humans to perform medical claim settlement audits. In the field of financial technology, similar problems also exist, making it more difficult for humans to perform financial claim settlement audits.

[0009] Therefore, how to provide a claim settlement exception identification method, device, computer equipment and computer readable storage medium, which can effectively improve the accuracy and efficiency of claim settlement exception identification, is a problem that needs to be solved by the technical personnel in the field at present. SUMMARY

[0010] In view of the deficiencies of the prior art described above, the purpose of the present application is to provide a claim settlement exception identification method, device, computer equipment and computer readable storage medium, aiming to solve the problem of how to effectively improve the accuracy and efficiency of claim settlement exception identification.

[0011] In order to achieve the above purpose, the present application adopts the following technical solutions:

[0012] In a first aspect, the present application provides a claim settlement exception identification method, comprising:

[0013] Obtaining case information of a claim settlement case uploaded by a target customer;

[0014] Extracting features of the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information;

[0015] Integrating the case information, the contradiction point detection result and historical claim settlement records of the target customer to obtain associated content of the claim settlement case, and generating an input text of a pre-trained claim settlement exception identification model according to the associated content;

[0016] Generating an exception identification result of the claim settlement case based on the input text through the claim settlement exception identification model.

[0017] In a second aspect, the present application provides a claim settlement exception identification device, comprising:

[0018] An information acquisition module for acquiring case information of a claim settlement case uploaded by a target customer;

[0019] A feature extraction module for extracting features of the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information;

[0020] A content integration module for integrating the case information, the contradiction point detection result and historical claim settlement records of the target customer to obtain associated content of the claim settlement case, and generating an input text of a pre-trained claim settlement exception identification model according to the associated content;

[0021] A result generation module for generating an exception identification result of the claim settlement case based on the input text through the claim settlement exception identification model.

[0022] In a third aspect, the present application provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the claim settlement exception identification method as described above when executing the computer program.

[0023] In a fourth aspect, the present application provides a computer readable storage medium storing a computer program, wherein the computer program implements the claim settlement exception identification method as described above when executed by a processor.

[0024] Compared with the prior art, the present application provides a claim settlement exception identification method, device, computer device and computer readable storage medium, wherein the case information of a claim settlement case uploaded by a target customer is obtained; feature extraction is performed on the case information to obtain a feature vector of the case information, and the feature vector is input into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information; the case information, the contradiction point detection result and the historical claim settlement record of the target customer are subjected to content integration processing to obtain associated content of the claim settlement case, and an input text of a pre-trained claim settlement exception identification model is generated according to the associated content; based on the input text, an exception identification result of the claim settlement case is generated by the claim settlement exception identification model; thereby the accuracy and efficiency of claim settlement exception identification can be effectively improved by the present application. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0026] Figure 1 An application environment schematic diagram of a claim settlement exception identification method provided by an embodiment of the present application.

[0027] Figure 2 A flowchart schematic diagram of a claim settlement exception identification method provided by an embodiment of the present application.

[0028] Figure 3 A program module schematic diagram of a claim settlement exception identification device provided by an embodiment of the present application.

[0029] Figure 4 A structure schematic diagram of a computer device provided by an embodiment of the present application.

[0030] Figure 5Another structural schematic diagram of a computer device provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0032] It should be understood that, when used in the specification and the appended claims of the present application, the term “comprising” indicates the presence of the described features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0033] It should also be understood that, when used in the specification and the appended claims of the present application, the term “and / or” refers to any combination of one or more of the associated listed items and all possible combinations thereof, and includes these combinations.

[0034] As used in the specification and the appended claims of the present application, the term “if” can be interpreted as “when” or “upon” or “in response to a determination” or “in response to detecting” depending on the context. Similarly, the phrase “if it is determined” or “if [a described condition or event] is detected” can be interpreted as meaning “upon determining” or “in response to determining” or “upon detecting [a described condition or event]” or “in response to detecting [a described condition or event]” depending on the context.

[0035] In addition, in the description of the present application and the appended claims, the terms “first”, “second”, “third”, and the like are only used for differentiation in description, and cannot be understood as indicating or implying relative importance.

[0036] In the present application, the reference to “one embodiment” or “some embodiments” and the like means that the specific features, structures, or characteristics described in connection with the embodiment are included in one or more embodiments of the present application. Therefore, the statements “in one embodiment”, “in some embodiments”, “in other some embodiments”, “in yet some embodiments”, and the like appearing in different places in the specification are not necessarily all referring to the same embodiment, but mean “one or more but not all embodiments”, unless otherwise specifically emphasized. The terms “comprise”, “include”, “have”, and their variations mean “including but not limited to”, unless otherwise specifically emphasized.

[0037] It should be understood that the size of the serial number of each step in the following embodiment does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0038] In order to illustrate the technical solutions of the present application, the following will be illustrated by specific embodiments.

[0039] An embodiment of the present application provides a claim settlement exception identification method, which can be applied in an application environment as shown in the figure. Figure 1 The client includes but is not limited to a palm computer, a desktop computer, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a cloud computer device, a personal digital assistant (PDA), etc. The server can be a stand-alone server, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content distribution networks (CDN), and big data and artificial intelligence platforms, etc. Basic cloud computing services.

[0040] Please refer to Figure 2 An embodiment of the present application provides a claim settlement exception identification method, which comprises the following steps:

[0041] S100, obtaining case information of a claim settlement case uploaded by a target customer;

[0042] S200, performing feature extraction on the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information;

[0043] S300, performing content integration processing on the case information, the contradiction point detection result and historical claim settlement records of the target customer to obtain associated content of the claim settlement case, and generating an input text of a pre-trained claim settlement exception identification model according to the associated content;

[0044] S400, generating an exception identification result of the claim settlement case based on the input text through the claim settlement exception identification model.

[0045] In specific implementation, the claim settlement exception identification method of the present embodiment can effectively improve the accuracy and efficiency of claim settlement exception identification through a series of innovative steps and technical means. The specific analysis is as follows:

[0046] 1. Data acquisition and feature extraction (S100 and S200)

[0047] Case information acquisition: First, acquire the case information of the target customer uploaded for the claim case, which includes multi-dimensional data such as accident description, claim amount, accident time, accident location, etc.

[0048] Feature extraction and contradiction point detection: Perform feature extraction on the case information, generate a feature vector, and input it into the pre-trained contradiction point detection model to generate a contradiction point detection result. This process not only extracts the key features of the case information, but also identifies potential contradictions in the case information through the contradiction point detection model, which are often important clues for abnormal behavior.

[0049] 2. Content integration and input text generation (S300)

[0050] Multi-data source fusion: Integrate the case information, contradiction point detection results, and target customer's historical claim records to generate the associated content of the claim case. By integrating multi-source data, the relevance between different data sources is fully utilized to provide more comprehensive information support for subsequent anomaly recognition.

[0051] Generate input text: Generate the input text of the pre-trained claim anomaly recognition model based on the integrated associated content. This process ensures the completeness and accuracy of the input text, providing high-quality input data for the claim anomaly recognition model.

[0052] 3. Anomaly recognition (S400)

[0053] Model inference: Input the generated input text into the pre-trained claim anomaly recognition model to generate the anomaly recognition result of the claim case. The claim anomaly recognition model is based on deep learning algorithms and can automatically learn and recognize abnormal patterns to provide high-accuracy anomaly recognition results.

[0054] Dynamic update and adaptive ability: The claim anomaly recognition model has online learning ability and can update model parameters in real time to adapt to new data and abnormal patterns, ensuring the long-term effectiveness and accuracy of the model.

[0055] Through the above steps, the following technical effects are achieved:

[0056] 1. Improve accuracy:

[0057] Contradiction point detection: Identify potential contradictions in the case information through the contradiction point detection model, providing important clues for anomaly recognition and significantly improving the accuracy of anomaly recognition.

[0058] Multi-data source fusion: Integrates case information, contradiction point detection results, and historical claim records, fully utilizes the relevance between different data sources, and further improves the accuracy of anomaly identification.

[0059] Model optimization: The pre-trained claim anomaly identification model combines online learning capabilities, which can update model parameters in real time, adapt to new data and anomaly patterns, and ensure long-term accuracy of the model.

[0060] 2. Improve efficiency:

[0061] Automatic processing: The entire process from data acquisition to anomaly identification is automated, reducing the workload and time cost of manual review.

[0062] Real-time update: The online learning capability of the claim anomaly identification model allows it to update in real time, quickly adapt to new data and anomaly patterns, and improve processing efficiency.

[0063] Result interpretation: Detailed anomaly identification result interpretation text can be generated to help claim personnel quickly understand the decision basis of the claim anomaly identification model, improving the efficiency of claim processing.

[0064] That is, the present application significantly improves the accuracy and efficiency of claim anomaly identification through the innovations of contradiction point detection, multi-data source fusion, dynamic model updating, and result interpretation, providing an efficient, accurate, and reliable claim anomaly identification method for the insurance industry.

[0065] It can be understood that the claim anomaly identification method provided by the embodiments of the present application can be applied to the claim anomaly identification scenarios related to the medical and health field. The following is a specific example:

[0066] Scenario description

[0067] In the medical and health field, insurance claim cases often involve a large amount of complex information such as medical records, diagnosis reports, and treatment cost details. These information not only has a large amount of data, but also is highly professional, making manual review extremely difficult. For example, a patient may have multiple medical records due to multiple hospitalizations, each record containing detailed diagnosis information, treatment process, and cost details.

[0068] Application of the claim anomaly identification method of the present application

[0069] 1. Obtain case information:

[0070] Obtain the patient's medical records from the hospital information system (HIS), including diagnosis reports, treatment processes, and cost details.

[0071] Obtain the patient's claim application information from the insurance company's claim system, including claim amount, accident time, and accident location.

[0072] 2. Feature extraction and contradiction detection:

[0073] For example, extract key features from medical records such as diagnosis codes, treatment items, cost amounts, etc.

[0074] Input the extracted feature vector into the pre-trained contradiction detection model to identify potential contradictions. For example, detect that the diagnosis code of a certain hospitalization does not match the treatment items, or that certain items in the cost details do not match the diagnosis result.

[0075] 3. Content integration and input text generation:

[0076] Integrate medical records, contradiction detection results, and patient's historical claim records, etc. to generate associated content for claim cases.

[0077] Generate input text for the pre-trained claim exception identification model based on the integrated associated content. The input text includes detailed medical records, contradiction descriptions, and historical claim information, etc.

[0078] 4. Exception identification:

[0079] Input the generated input text into the pre-trained claim exception identification model to generate the exception identification result of the claim case.

[0080] The claim exception identification model is based on deep learning algorithms to identify abnormal behaviors such as fraud, incorrect information, etc., and generate detailed explanation texts to help claim personnel quickly understand the model's decision basis.

[0081] Technical effects

[0082] Through the above steps, the present application can effectively identify abnormal behaviors in medical claims, improving the accuracy and efficiency of claim processing. For example, through the present application, it can be identified that there are repeated reimbursement items in the cost details of a certain hospitalization, or that the diagnosis does not match the actual treatment, thereby helping insurance companies to timely discover and handle potential fraudulent behaviors.

[0083] It can be understood that the claim exception identification method provided by the embodiments of the present application can also be applied to the claim exception identification scenarios related to the field of financial technology. The following is a specific example:

[0084] Scenario description

[0085] In the field of financial technology, the processing of insurance claim cases often involves a large amount of digitized data, including online claim applications, electronic contracts, payment records, blockchain technology, etc. These data not only need to be processed quickly, but also need to ensure the security and privacy of the data. For example, a customer may submit a claim application through a mobile application while uploading relevant electronic contracts and payment records.

[0086] Application of the claim exception identification method

[0087] Suppose a customer submits a property insurance claim application through a financial technology platform, with a claim amount of 100,000 yuan and an accident time of July 15, 2025. The insurance company processes it through the following steps:

[0088] 1. Obtain case information:

[0089] Obtain the customer's claim application information from the financial technology platform's claim system, including the claim amount of 100,000 yuan, the accident time of July 15, 2025, the accident description, the property loss details, etc.

[0090] Obtain the customer's electronic contract information from the electronic contract management system, including the insurance terms, the insurance amount, the insurance period, etc.

[0091] Obtain the customer's payment records from the payment system, including premium payment records, claim payment records, etc.

[0092] 2. Feature extraction and contradiction detection:

[0093] Extract key features from the claim application, electronic contract and payment records, such as claim amount, accident time, insurance amount, premium payment status, etc.

[0094] Input the extracted feature vector into the pre-trained contradiction detection model to identify potential contradictions. For example, it is detected that the claim amount in the claim application exceeds the insurance amount in the electronic contract, or the premium payment record shows that the customer has not paid the premium on time.

[0095] 3. Content integration and input text generation:

[0096] Integrate the claim application information, contradiction detection results and customer's historical claim records, etc. to generate the associated content of the claim case.

[0097] Generate the input text of the pre-trained claim exception identification model based on the integrated associated content. The input text includes detailed claim application information, contradiction description and historical claim information, etc.

[0098] 4. Exception identification:

[0099] The generated input text is input into the pre-trained claim exception identification model to generate an exception identification result of the claim case.

[0100] The claim exception identification model is based on a deep learning algorithm and identifies abnormal behaviors such as fraud, incorrect information, etc., and generates detailed explanation text to help claim personnel quickly understand the basis for the model's decision.

[0101] Technical effects

[0102] Through the above steps, the present application can effectively identify abnormal behaviors in the field of financial technology, improve the accuracy and efficiency of claim processing. For example, through the present application, it can identify the amount exceeding the limit in the claim application or the abnormal state of premium payment, so as to help the insurance company to discover and handle potential fraudulent behaviors in time. At the same time, combined with the blockchain technology, the safety and privacy protection of data are ensured, and the reliability and credibility of claim processing are further improved.

[0103] Further, in one embodiment, the claim exception identification method, wherein the case information of the claim case uploaded by the target customer is obtained, specifically comprising the steps of:

[0104] Receiving the case information of the claim case uploaded by the target customer through a pre-set data transmission interface;

[0105] Pretreating the case information by removing duplicate data, correcting format errors, and filling in missing fields;

[0106] Verifying the pretreated case information to check whether it meets the data integrity requirements, and triggering an alarm and notifying the target customer to supplement or correct the case information that does not meet the data integrity requirements.

[0107] In specific implementation, the specific implementation process of the steps of the present embodiment is approximately as follows:

[0108] 1. Receiving case information through a pre-set data transmission interface

[0109] 11. Interface configuration:

[0110] Configure a secure data transmission interface (such as RESTful API or SOAP API) to ensure that the interface can receive data in multiple formats (such as JSON, XML, CSV, etc.).

[0111] Set up an identity verification mechanism (such as OAuth2.0 or API key) to ensure that only authorized users can upload data.

[0112] 12. Receive data:

[0113] Listen to the interface, receive the target customer uploaded case information of the claim case.

[0114] The received data is analyzed and the basic content of the case information (such as claim amount, accident time, accident location, etc.) is extracted.

[0115] 13. Log:

[0116] Record the timestamp, data format, data size, etc. of each data reception, which is convenient for subsequent audit and problem troubleshooting.

[0117] 2. Pre-process the case information

[0118] 21. Remove duplicate data:

[0119] Use a hash algorithm (such as SHA-256) to generate a unique identifier for the key fields in the case information (such as claim number, accident time, accident location, etc.).

[0120] Compare the existing case information in the database to identify and remove duplicate records.

[0121] 22. Correct format errors:

[0122] Regular expression matching is performed on the text field to correct common format errors (such as date format, phone number format, etc.).

[0123] Range check is performed on the numerical field to ensure that the data is within a reasonable range.

[0124] 23. Fill in missing fields:

[0125] For missing fields, they can be filled in according to business rules.

[0126] For fields that cannot be filled in, record the missing information for subsequent processing.

[0127] 3. Data verification on the pre-processed case information

[0128] 31. Data integrity check:

[0129] Check if the case information meets the data integrity requirements, including whether all mandatory fields exist, and whether the field values meet the business logic, etc.

[0130] For example, check if the claim amount is positive, and if the accident time is within the insurance validity period, etc.

[0131] 32. Trigger alarm:

[0132] For case information that does not meet the data integrity requirements, trigger the alarm mechanism. The alarm can be sent through email, SMS or system notification.

[0133] Detailed information of the alert is recorded, including case number, non-compliant fields, alert time, etc.

[0134] 33. Notify the customer:

[0135] Through the preset notification channel (such as email, SMS, mobile application push, etc.), notify the target customer to supplement or correct the data.

[0136] Provide clear guidance information to inform the customer of the specific content and operation steps that need to be corrected.

[0137] 34. Log:

[0138] Record the sending time, recipient, notification content, etc. of each notification, for subsequent audit and problem troubleshooting.

[0139] Through the above process, the embodiment can realize efficient receiving, preprocessing, verification and notification of the claim case information uploaded by the target customer. These steps not only ensure the integrity and accuracy of the data, but also improve the efficiency and reliability of data processing through automated processes.

[0140] Further, in one embodiment, the claim exception identification method, wherein the feature extraction of the case information obtains the feature vector of the case information, and the feature vector is input into the pre-trained contradiction point detection model to generate the contradiction point detection result of the case information, specifically including steps:

[0141] Load the pre-trained contradiction point detection model;

[0142] Extract the text features, numerical features, category features and time series features of the case information;

[0143] Standardize the text features, numerical features, category features and time series features;

[0144] Fuse the standardized text features, numerical features, category features and time series features to obtain the feature vector of the case information;

[0145] Input the feature vector into the contradiction point detection model to generate the contradiction point detection result of the case information.

[0146] In specific implementation, the specific implementation process of the steps of the embodiment is approximately as follows:

[0147] 1. Load the pre-trained contradiction point detection model

[0148] 11. Model loading:

[0149] Load the contradiction detection model from the pre-trained model repository. This model is usually a deep learning model (such as BERT, BiLSTM, etc.) that can identify contradictions in case information.

[0150] Ensure that the model has been loaded into memory so that it can quickly respond to subsequent feature vector inputs.

[0151] 12. Model Validation:

[0152] Verify the integrity and validity of the model to ensure that the model parameters are loaded correctly and the model can run normally.

[0153] The model can be quickly verified using the first test data set to ensure its accuracy and stability.

[0154] 2. Extract features of case information

[0155] 21. Text Feature Extraction:

[0156] Use natural language processing (NLP) technology to process text fields in case information (such as accident description, claim reason, etc.).

[0157] Extract features such as keywords, part-of-speech tags, and text vectors. For example, use TF-IDF or Word2Vec to convert text into vector form.

[0158] 22. Numerical feature extraction:

[0159] Extract numerical fields from case information (such as claim amount, accident time, claim frequency, etc.).

[0160] Normalize numeric fields to ensure consistent value ranges.

[0161] 23. Category feature extraction:

[0162] Extract category fields from case information (such as accident type, claim type, customer type, etc.).

[0163] One-Hot Encoding is performed on the category field to convert the category features into numerical form.

[0164] 24. Time Series Feature Extraction:

[0165] Extract time series fields from case information (such as claim time, accident time, etc.).

[0166] Normalize the time series field into a time window and extract the statistical characteristics of the time series (such as mean, variance, maximum, minimum, etc.).

[0167] 3. Feature Standardization

[0168] 31. Numerical Feature Standardization:

[0169] Z-score standardization is applied to numerical features, converting them to a distribution with a mean of 0 and a standard deviation of 1.

[0170] The formula is: normalized value = (original value - mean) / standard deviation

[0171] 32. Text Feature Normalization:

[0172] Normalization is applied to text features to ensure consistent length of text vectors.

[0173] For example, L2 norm is used to normalize text vectors.

[0174] 33. Category Feature One-Hot Encoding:

[0175] One-hot encoding is applied to category features, converting them to numerical form.

[0176] Ensure that the dimensionality of the one-hot encoded features is consistent for subsequent processing.

[0177] 34. Time Series Feature Normalization:

[0178] Normalization is applied to time series features to ensure consistent range of time series.

[0179] For example, normalize time series to the [0, 1] interval.

[0180] 4. Feature Fusion

[0181] 41. Feature Concatenation:

[0182] Concatenate the normalized text features, numerical features, category features, and time series features to form a complete feature vector.

[0183] Ensure that the dimensionality of the feature vector is consistent for input into the contradiction point detection model.

[0184] 42. Feature Weighting:

[0185] According to the importance and relevance of features, assign different weights to each feature in the feature vector.

[0186] For example, assign higher weights to high-risk numerical features (such as claim amounts).

[0187] 43. Feature Dimensionality Reduction:

[0188] Using Principal Component Analysis (PCA) or other dimensionality reduction techniques to reduce the dimensionality of the feature vectors and improve the efficiency of the contradiction point detection model.

[0189] Ensure that the feature vectors after dimensionality reduction can retain enough information and do not affect the performance of the contradiction point detection model.

[0190] 5. Input the contradiction point detection model and generate results

[0191] 51. Model input:

[0192] Input the fused feature vectors into the pre-trained contradiction point detection model.

[0193] Ensure that the format and dimension of the feature vectors meet the requirements of the contradiction point detection model.

[0194] 52. Model inference:

[0195] The contradiction point detection model infers the input feature vectors and generates contradiction point detection results.

[0196] The detection results may include specific descriptions of the contradiction points, confidence of the contradiction points, etc.

[0197] 53. Result processing:

[0198] Process the contradiction point detection results generated by the contradiction point detection model to extract key information.

[0199] For example, extract the specific location of the contradiction point, the type of the contradiction point, the confidence of the contradiction point, etc.

[0200] 54. Result storage and notification:

[0201] Store the contradiction point detection results in the database for subsequent analysis and processing.

[0202] If a contradiction point is detected, an alarm can be triggered and relevant staff can be notified for further investigation.

[0203] Through the above process, the embodiment can realize feature extraction, standardization processing, feature fusion and contradiction point detection of case information. These steps not only ensure the high quality and consistency of the feature vectors, but also efficiently identify the contradiction points in the case information through the pre-trained contradiction point detection model, providing an important basis for subsequent claim abnormality identification.

[0204] Further, in one embodiment, the claim abnormality identification method, wherein the content integration processing of the case information, the contradiction point detection result and the historical claim record of the target customer obtains the associated content of the claim case, and the input text of the pre-trained claim abnormality identification model is generated according to the associated content, specifically including the following steps:

[0205] store the case information, the contradiction point detection result and the historical claim record of the target customer in a graph structure using graph database technology, and generate the associated content of the claim case;

[0206] adjust the content and format of the associated content using context awareness technology;

[0207] generate a prompt text of a pre-trained claim exception recognition model according to a preset prompt word generation strategy based on the adjusted associated content;

[0208] match the prompt text with the input format of the claim exception recognition model, and if the format matching is passed, use the prompt text as the input text of the claim exception recognition model.

[0209] Further, the claim exception recognition method, wherein the storing the case information, the contradiction point detection result and the historical claim record of the target customer in a graph structure using graph database technology, and generating the associated content of the claim case, specifically comprises the steps of:

[0210] constructing the structure of nodes and edges of the graph database according to the features of the case information, the contradiction point detection result and the historical claim record;

[0211] based on the constructed structure of nodes and edges of the graph database, importing the case information, the contradiction point detection result and the historical claim record into the graph database for storage;

[0212] based on the data in the graph database, querying through a graph query language to generate the associated content of the claim case.

[0213] Further, the claim exception recognition method, wherein the matching the prompt text with the input format of the claim exception recognition model, and if the format matching is passed, using the prompt text as the input text of the claim exception recognition model, further comprises the steps of:

[0214] if the format matching is not passed, adjusting the format of the prompt text;

[0215] matching the format-adjusted prompt text with the input format of the claim exception recognition model again until the format matching is passed.

[0216] In specific implementation, the specific implementation process of the steps of the present embodiment is as follows:

[0217] 1. Content integration using graph database technology

[0218] 11. Constructing the graph database structure:

[0219] Node design: Design the types of nodes in the graph database based on the characteristics of case information, contradiction point detection results, and historical claim records. For example:

[0220] Claim case node: Contains attributes such as case number, claim amount, accident time, and accident location.

[0221] Contradiction point node: Contains attributes such as contradiction point description, contradiction point type, and confidence.

[0222] Customer node: Contains attributes such as customer number, customer name, and customer type.

[0223] Historical claim record node: Contains attributes such as claim number, claim amount, and claim time.

[0224] Edge design: Define the relationships between nodes. For example:

[0225] Relationship between claim case and customer: Indicates that the claim case belongs to a certain customer.

[0226] Relationship between claim case and contradiction point: Indicates that there is a certain contradiction point in the claim case.

[0227] Relationship between claim case and historical claim record: Indicates the association between the claim case and the historical claim record.

[0228] 12. Data import and storage:

[0229] Data parsing: Parse case information, contradiction point detection results, and historical claim records to extract key information.

[0230] Data import: Import the parsed data into the graph database and store it according to the designed node and edge structure.

[0231] Data verification: Ensure the integrity and consistency of the imported data and handle errors that may occur during data import.

[0232] 13. Generate associated content:

[0233] Graph query: Use a graph query language (such as Cypher) to perform complex queries and extract the association between case information, contradiction point detection results, and historical claim records.

[0234] Associated content generation: Integrate the query results into structured associated content, such as generating a JSON object containing detailed information about the claim case, contradiction point description, and historical claim records.

[0235] 2. Use context-aware technology to adjust content and format

[0236] 21. Content Adjustment:

[0237] Semantic Analysis: Perform semantic analysis on the generated associated content to identify key information and logical relationships.

[0238] Content Optimization: Based on the results of semantic analysis, optimize the expression of associated content to ensure accuracy and readability. For example, adjust the logical structure of the text and supplement missing information.

[0239] 22. Format Adjustment:

[0240] Format Standardization: Adjust the format of associated content to a unified standard format, such as unified date format, amount format, etc.

[0241] Format Verification: Verify whether the adjusted format meets the preset format requirements to ensure consistency.

[0242] 3. Generate Prompt Text

[0243] 31. Prompt Word Generation Strategy:

[0244] Strategy Definition: Define the generation rules of prompt text according to the preset prompt word generation strategy. For example, generate different prompt words according to the severity of the claim case and the type of conflict points.

[0245] Prompt Text Generation: Generate prompt text for the pre-trained claim exception identification model based on the adjusted associated content. The prompt text should contain key information and logical relationships to facilitate the understanding and processing of the claim exception identification model.

[0246] 32. Text Optimization:

[0247] Text Polishing: Polish the generated prompt text to ensure text fluency and readability.

[0248] Content Integrity Check: Ensure that the prompt text contains all necessary information to avoid information omission.

[0249] 4. Input Format Matching

[0250] 41. Format Matching:

[0251] Format Check: Match the generated prompt text with the input format of the claim exception identification model to check whether it meets the requirements of the claim exception identification model.

[0252] Matching Result Processing:

[0253] If the matching is successful: Take the prompt text as the input text of the claim exception identification model and proceed to the next step.

[0254] If the match fails: Go to the format adjustment step.

[0255] 42. Format adjustment:

[0256] Adjustment strategy: According to the matching result, determine the format content that needs to be adjusted. For example, adjust the length of the text, the order of the fields, etc.

[0257] Adjustment operation: Adjust the format of the prompt text to ensure that the adjusted text meets the input format requirements of the claim exception recognition model.

[0258] 43. Match again:

[0259] Repeat the matching: Adjust the prompt text again and match it with the input format of the claim exception recognition model until the format matching is passed.

[0260] Record the adjustment process: Record the details of each adjustment for subsequent audit and problem troubleshooting.

[0261] Through the above process, this embodiment can realize efficient integration and processing of case information, conflict point detection results and historical claim records, and generate input text that meets the input requirements of the claim exception recognition model. These steps not only ensure the integrity and consistency of the data, but also improve the accuracy and reliability of data processing through context perception technology and format matching mechanism.

[0262] Further, in one embodiment, the claim exception recognition method, wherein the generating the claim case exception recognition result based on the input text through the claim exception recognition model comprises the steps of:

[0263] Loading the claim exception recognition model;

[0264] Inputting the input text into the claim exception recognition model to generate a preliminary recognition result of the claim case;

[0265] Formatting the preliminary recognition result to obtain the claim case exception recognition result, and outputting or displaying the claim case exception recognition result.

[0266] In specific implementation, the specific implementation process of the steps of the embodiment is as follows:

[0267] 1. Load the claim exception recognition model

[0268] 11. Model loading:

[0269] Load the claim exception recognition model from the pre-trained model repository. The model is usually a deep learning model (such as BERT, BiLSTM, etc.), which can identify abnormal behaviors in claim cases.

[0270] Ensure that the claim exception identification model has been loaded into memory for quick response to subsequent input text.

[0271] 12. Model verification:

[0272] Verify the integrity and effectiveness of the model, ensuring that the model parameters are correctly loaded and the model can run normally.

[0273] The model can be quickly verified by the second test data set to ensure the accuracy and stability of the model.

[0274] 13. Model initialization:

[0275] Initialize the running environment of the model, including setting necessary hyperparameters (such as batch size, maximum sequence length, etc.).

[0276] Ensure that the model is ready to receive input text and perform inference.

[0277] 2. Preliminary identification of input text

[0278] 21. Model inference:

[0279] Input the input text into the claim exception identification model to generate preliminary identification results.

[0280] The claim exception identification model performs inference on the input text based on deep learning algorithms to identify possible abnormal behaviors (such as fraud, incorrect information, etc.).

[0281] The preliminary identification results may include the type of abnormal behavior, confidence, etc.

[0282] 22. Result storage:

[0283] Store the preliminary identification results in a temporary storage area for subsequent formatting processing.

[0284] 3. Formatting processing of preliminary identification results

[0285] 31. Result analysis:

[0286] Analyze the preliminary identification results and extract key information such as the type of abnormal behavior, confidence, and involved fields.

[0287] Ensure that the parsed results structure is clear, making it easier for subsequent formatting processing.

[0288] 32. Formatting processing:

[0289] Format the preliminary identification results, which may include:

[0290] Convert the results to a unified JSON format.

[0291] Normalize the confidence score to ensure it is within the range [0, 1].

[0292] Text polish the description of abnormal behavior to ensure the readability of the results.

[0293] 4. Output or display of abnormal identification results

[0294] 41. Result output:

[0295] Output the formatted abnormal identification results to the designated system or interface. The output methods may include:

[0296] Store the results in the database for subsequent analysis and processing.

[0297] Return the results to the front-end system through the API interface for users to view.

[0298] Save the results in file form to the specified path for offline analysis.

[0299] 42. Result display:

[0300] If the results need to be displayed, design a user-friendly interface (such as a Web interface, mobile application interface, etc.) to display the abnormal identification results to users in an intuitive way.

[0301] The display content may include detailed description of abnormal behavior, confidence, suggested handling measures, etc.

[0302] 43. Notification and alarm:

[0303] For high-confidence abnormal behavior, trigger the alarm mechanism to notify relevant staff for further investigation and handling. Notification methods may include email, SMS, system notification, etc.

[0304] Through the above process, this embodiment can realize efficient generation, formatting processing and output display of abnormal identification results of claim cases. These steps not only ensure the accuracy and consistency of the results, but also improve the practicality and operability of claim abnormal identification through user-friendly display and notification mechanism.

[0305] From the above method embodiment, the claim settlement exception identification method provided by the application comprises: obtaining case information of a claim settlement case uploaded by a target customer; performing feature extraction on the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information; performing content integration processing on the case information, the contradiction point detection result and historical claim settlement records of the target customer to obtain associated content of the claim settlement case, and generating input text of a pre-trained claim settlement exception identification model according to the associated content; and generating an exception identification result of the claim settlement case by the claim settlement exception identification model based on the input text. In this way, the method of the application can effectively improve the accuracy and efficiency of claim settlement exception identification.

[0306] It should be understood that although the present application provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps can be included based on conventional or non-inventive labor, and the operation steps are not necessarily executed in the order of the embodiments or flowcharts. The order of steps listed in the embodiments or flowcharts is only one of the many execution orders, and does not represent the only execution order. It should be noted that there is no certain sequence between the above steps, and those skilled in the art can understand from the description of the embodiments of the present application that the above steps can have different execution orders in different embodiments, i.e., they can be executed in parallel, or exchanged for execution, etc. Moreover, at least part of the steps in the embodiments or flowcharts can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily sequential, but can be executed in rotation, alternation or synchronization with other steps or sub-steps or stages of other steps.

[0307] Based on the above method embodiment, please refer to Figure 3 The other embodiment of the present application further provides a claim settlement exception identification device, wherein the device comprises:

[0308] The information acquisition module 11 is configured to acquire case information of a claim settlement case uploaded by a target customer.

[0309] The feature extraction module 12 is configured to perform feature extraction on the case information to obtain a feature vector of the case information, and input the feature vector into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information.

[0310] The content integration module 13 is configured to integrate the case information, the contradiction point detection result and the historical claim record of the target customer to obtain associated content of the claim case, and generate input text of the pre-trained claim exception identification model according to the associated content.

[0311] The result generation module 14 is configured to generate an exception identification result of the claim case by the claim exception identification model based on the input text.

[0312] Further, in an embodiment, the claim exception identification device, wherein the case information uploaded by the target customer is obtained, and specifically includes:

[0313] The case information uploaded by the target customer is received through a preset data transmission interface.

[0314] The case information is preprocessed by removing duplicate data, correcting format errors and filling in missing fields.

[0315] The preprocessed case information is verified to check whether it meets the data integrity requirement, and if the case information does not meet the data integrity requirement, an alarm is triggered and the target customer is notified to supplement or correct.

[0316] Further, in an embodiment, the claim exception identification device, wherein the feature extraction is performed on the case information to obtain a feature vector of the case information, and the feature vector is input into a pre-trained contradiction point detection model to generate a contradiction point detection result of the case information, and specifically includes:

[0317] The pre-trained contradiction point detection model is loaded.

[0318] Text features, numerical features, category features and time sequence features of the case information are extracted.

[0319] The text features, the numerical features, the category features and the time sequence features are standardized.

[0320] The standardized text features, the numerical features, the category features and the time sequence features are fused to obtain the feature vector of the case information.

[0321] The feature vector is input into the contradiction point detection model to generate the contradiction point detection result of the case information.

[0322] Further, in one embodiment, the claim settlement exception identification device, wherein the content integration processing of the case information, the contradiction point detection result and the historical claim settlement record of the target customer is to obtain the associated content of the claim settlement case, and the input text of the pre-trained claim settlement exception identification model is generated according to the associated content, specifically comprising:

[0323] The case information, the contradiction point detection result and the historical claim settlement record of the target customer are stored in the form of a graph structure by using a graph database technology, and the associated content of the claim settlement case is generated;

[0324] The associated content is adjusted in content and format by using a context perception technology;

[0325] The prompt text of the pre-trained claim settlement exception identification model is generated according to a preset prompt word generation strategy based on the adjusted associated content;

[0326] The prompt text is matched with the input format of the claim settlement exception identification model, and if the format matching passes, the prompt text is taken as the input text of the claim settlement exception identification model.

[0327] Further, the claim settlement exception identification device, wherein the case information, the contradiction point detection result and the historical claim settlement record of the target customer are stored in the form of a graph structure by using a graph database technology, and the associated content of the claim settlement case is generated, specifically comprising:

[0328] According to the characteristics of the case information, the contradiction point detection result and the historical claim settlement record, the structure of the nodes and edges of the graph database is constructed;

[0329] Based on the constructed structure of the nodes and edges of the graph database, the case information, the contradiction point detection result and the historical claim settlement record are imported into the graph database for storage;

[0330] Based on the data in the graph database, the associated content of the claim settlement case is generated by querying through a graph query language.

[0331] Further, the claim settlement exception identification device, wherein the prompt text is matched with the input format of the claim settlement exception identification model, and if the format matching passes, the prompt text is taken as the input text of the claim settlement exception identification model, and specifically comprising:

[0332] If the format matching does not pass, the prompt text is adjusted in format;

[0333] The prompt text after format adjustment is matched with the input format of the claim exception identification model again until the format matching is passed.

[0334] Further, in one embodiment, the claim exception identification device, wherein the abnormal identification result of the claim case is generated by the claim exception identification model based on the input text, specifically comprising:

[0335] Load the claim exception identification model;

[0336] Input the input text into the claim exception identification model to generate a preliminary identification result of the claim case;

[0337] Format the preliminary identification result to obtain the abnormal identification result of the claim case, and output or display the abnormal identification result.

[0338] It should be noted that the information interaction, execution process, etc. between the above-mentioned modules in the device embodiment of the application are based on the same concept as the method embodiment of the application, and the specific functions and technical effects brought by them can be referred to the method embodiment part, which will not be repeated here.

[0339] Based on the above method embodiment, another embodiment of the application further provides a computer device, which can be a server, and the internal structure diagram thereof can be as shown in Figure 4 The computer device includes a processor, a memory, a network interface and a database connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with the external terminal through the network connection. The computer program is executed by the processor to realize the functions or steps of the claim exception identification method server side in any one of the above method embodiments.

[0340] Based on the above method embodiment, another embodiment of the application further provides a computer device, which can be a client, and the internal structure diagram thereof can be as shown in Figure 5As shown in the structural schematic diagram. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the computer device is used to communicate with the external terminal through the network connection. The computer program is executed by the processor to realize the function or step of the claim settlement exception identification method client side in any one of the above method embodiments.

[0341] Those skilled in the art can understand that, Figure 4 With Figure 5 The structural schematic diagram shown in the figure is only a schematic diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. A specific computer device can include more components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0342] Among them, the so-called processor can be a CPU, and the processor can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), ready programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0343] The memory includes a readable storage medium, an internal memory, etc., wherein the internal memory can be a memory of the computer device, and the internal memory provides an environment for running the operating system and the computer readable instructions in the readable storage medium. The readable storage medium can be a hard disk of the computer device, and in other embodiments, can also be an external storage device of the computer device, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory can include both an internal storage unit of the computer device and an external storage device. The memory is used to store an operating system, an application program, a boot loader, data, and other programs, such as program codes of computer programs, etc. The memory can also be used to temporarily store data that has been output or will be output.

[0344] Based on the above method embodiments, another embodiment of the present application further provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the claim settlement exception identification method in any one of the above method embodiments. The computer readable storage medium can be non-volatile or volatile.

[0345] It should be noted that the functions or steps that the computer readable storage medium or the computer device can achieve and the technical effects brought by the functions / steps can be referred to the related description in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0346] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, storage, database or other medium used in each embodiment provided in the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc. The disclosed memory components or memories of the operating environment described herein are intended to include one or more of these and / or any other suitable type of memory.

[0347] Those skilled in the art can clearly understand that, for the convenience and brevity of description, in the device embodiment of the present application, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the above-described functions. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or software. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the unit and module in the above-mentioned device can refer to the corresponding process in the above-mentioned method embodiment, which will not be repeated here. If the integrated unit is realized in the form of software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium.

[0348] Those skilled in the art can understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0349] In the embodiments provided by the present application, it should be understood that the disclosed apparatus / computer device and method can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely schematic. The division of the modules or units is merely a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0350] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0351] It should be noted that if non-company software tools or components appear in the embodiments of the present application, they are only used for example introduction and do not represent actual use. The above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for identifying abnormal claims, characterized in that: include: Obtain case information of claims uploaded by target customers; Extracting features from the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction detection model to generate a contradiction detection result for the case information; Integrate the case information, the conflict point detection results, and the target customer's historical claims records to obtain relevant content of the claims case, and generate input text for a pre-trained claims anomaly recognition model based on the relevant content; Based on the input text, an anomaly recognition result of the claim case is generated by the claim anomaly recognition model.

2. The method for identifying claim anomalies according to claim 1, characterized in that: The case information of the claim case uploaded by the target customer is obtained, including: Receive the case information of claims uploaded by target customers through the preset data transmission interface; Pre-processing the case information to remove duplicate data, correct formatting errors, and fill in missing fields; The pre-processed case information is subjected to data verification to check whether it meets the data integrity requirements. For the case information that does not meet the data integrity requirements, an alarm is triggered and the target customer is notified to supplement or correct it.

3. The method for identifying claim anomalies according to claim 1, characterized in that: The extracting features of the case information to obtain a feature vector of the case information, and inputting the feature vector into a pre-trained contradiction detection model to generate a contradiction detection result of the case information includes: Load the pre-trained contradiction detection model; Extracting text features, numerical features, category features, and time series features of the case information; performing standardization processing on the text features, the numerical features, the category features, and the time series features; fusing the standardized text features, the numerical features, the category features, and the time series features to obtain a feature vector of the case information; The feature vector is input into the contradiction detection model to generate a contradiction detection result of the case information.

4. The method for identifying claim anomalies according to claim 1, characterized in that: The process of integrating the case information, the conflict point detection results, and the target customer's historical claims records to obtain relevant content of the claims case, and generating input text for a pre-trained claims anomaly recognition model based on the relevant content, includes: Using graph database technology, the case information, the conflict point detection results, and the target customer's historical claims records are stored in the form of a graph structure, and related content of the claim case is generated; Using context-aware technology, adjusting the content and format of the associated content; Based on the adjusted associated content, generating prompt text for the pre-trained claim anomaly recognition model according to a preset prompt word generation strategy; The prompt text is matched with the input format of the claim anomaly recognition model. If the format matching is successful, the prompt text is used as the input text of the claim anomaly recognition model.

5. The method for identifying claim anomalies according to claim 4, characterized in that: The graph database technology is used to store the case information, the conflict point detection results, and the target customer's historical claims records in the form of a graph structure, and to generate related content of the claim case, including: Constructing a node and edge structure of a graph database based on the case information, the conflict point detection results, and the characteristics of the historical claims records; Based on the constructed node and edge structure of the graph database, the case information, the conflict point detection results, and the historical claims records are imported into the graph database for storage; Based on the data in the graph database, a query is performed using a graph query language to generate related content of the claim case.

6. The method for identifying claim anomalies according to claim 4, characterized in that: The step of matching the prompt text with the input format of the claim anomaly recognition model and using the prompt text as the input text of the claim anomaly recognition model if the format matches, further includes: If the format matching fails, the format of the prompt text is adjusted; The format-adjusted prompt text is matched again with the input format of the claim anomaly recognition model until the format matching is successful.

7. The method for identifying claim anomalies according to claim 1, characterized in that: Generating an anomaly recognition result of the claim case based on the input text by the claim anomaly recognition model includes: Loading the claims anomaly recognition model; Inputting the input text into the claim anomaly recognition model to generate a preliminary recognition result of the claim case; The preliminary recognition result is formatted to obtain an abnormal recognition result of the claim case, and the abnormal recognition result is output or displayed.

8. A claim anomaly identification device, characterized in that: include: The information acquisition module is used to obtain the case information of the claim cases uploaded by the target customers; a feature extraction module, configured to extract features from the case information to obtain a feature vector of the case information, and input the feature vector into a pre-trained contradiction detection model to generate a contradiction detection result for the case information; a content integration module for integrating the case information, the conflict point detection results, and the target customer's historical claims records to obtain relevant content of the claim case, and generating input text for a pre-trained claims anomaly recognition model based on the relevant content; A result generation module is used to generate an anomaly recognition result of the claim case based on the input text through the claim anomaly recognition model.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for identifying claim anomalies according to any one of claims 1 to 7 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 identifying claim anomalies according to any one of claims 1 to 7 is implemented.

Images