Credit evaluation processing method and device

Abstract

The embodiment of the invention provides a credit evaluation processing method and device, and the method comprises the steps: inputting a to-be-evaluated image into a text extraction model for text extraction in a credit evaluation processing process, obtaining a structured text, carrying out the identity verification of the user identity information contained in the structured text, and obtaining a credit evaluation result; and performing similarity verification on the entity field contained in the structured text and a preset entity field, if the verification is passed, inputting the to-be-evaluated image and the prompt text into a large language model for verification processing, and after a verification result is obtained, synchronizing the structured text and the verification result to a credit evaluation platform.

Description

Technical Field

[0001] This document relates to the field of data processing technologies, and in particular, to a credit assessment processing method and apparatus. Background Art

[0002] With the development of Internet technologies, computers and the Internet have become important parts of users' lives and have gradually penetrated into every corner of life. Therefore, more and more online services have emerged. For example, in the credit assessment scenario, instead of using manual verification of assessment materials in the past, online credit assessment platforms and large models are used for credit assessment. In this process, how to better meet the needs of users and improve the accuracy and effectiveness of credit assessment has become the focus of increasing attention in the credit assessment process. Summary of the Invention

[0003] One or more embodiments of this specification provide a credit assessment processing method, including: inputting an image to be evaluated into a text extraction model for text extraction to obtain structured text; performing identity verification on the user identity information included in the structured text, and performing similarity verification on the entity fields included in the structured text and preset entity fields; if the verification passes, inputting the image to be evaluated and a prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: performing image feature extraction and feature fusion for each authenticity verification dimension, and determining the authenticity verification result by calculating an authenticity verification score for the fused features; synchronizing the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing.

[0004] One or more embodiments of this specification provide a credit assessment processing apparatus, including: a text extraction module configured to input an image to be evaluated into a text extraction model for text extraction to obtain structured text; a verification module configured to perform identity verification on the user identity information included in the structured text, and perform similarity verification on the entity fields included in the structured text and preset entity fields; if the verification passes, run an authenticity verification processing module, the authenticity verification processing module being configured to input the image to be evaluated and a prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: performing image feature extraction and feature fusion for each authenticity verification dimension, and determining the authenticity verification result by calculating an authenticity verification score for the fused features; a credit assessment module configured to synchronize the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing.

[0005] One or more embodiments of this specification provide a credit assessment processing device, including: a processor; and a memory configured to store computer-executable instructions, which when executed cause the processor to: input an image to be evaluated into a text extraction model for text extraction to obtain structured text. Verify the user identity information included in the structured text, and verify the similarity between the entity fields included in the structured text and preset entity fields. If the verification passes, input the image to be evaluated and a prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: extracting and fusing image features in each authenticity verification dimension, and determining the authenticity verification result by calculating an authenticity verification score for the fused features. Synchronize the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing.

[0006] One or more embodiments of this specification provide a computer-readable storage medium for storing computer-executable instructions, which when executed implement the following process: input an image to be evaluated into a text extraction model for text extraction to obtain structured text. Verify the user identity information included in the structured text, and verify the similarity between the entity fields included in the structured text and preset entity fields. If the verification passes, input the image to be evaluated and a prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: extracting and fusing image features in each authenticity verification dimension, and determining the authenticity verification result by calculating an authenticity verification score for the fused features. Synchronize the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings; Figure 1 It is a schematic diagram of the implementation environment of a credit assessment processing method provided by one or more embodiments of this specification; Figure 2 It is a processing flow chart of a credit assessment processing method provided by one or more embodiments of this specification; Figure 3 It is a processing flow chart of a credit assessment processing method applied to a credit limit assessment processing scenario provided by one or more embodiments of this specification; Figure 4Schematic diagram of an embodiment of a credit assessment processing device provided for one or more embodiments of this specification; Figure 5 Structural schematic diagram of a credit assessment processing device provided for one or more embodiments of this specification. Detailed implementation manners

[0008] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will clearly and completely describe the technical solutions in one or more embodiments of this specification with reference to the accompanying drawings in one or more embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of this document.

[0009] The credit assessment processing method provided by one or more embodiments of this specification is applicable to an implementation environment of a credit assessment system. Refer to Figure 1 , this implementation environment at least includes: Server 101, text extraction model 102, credit assessment platform 103, large language model 104. In addition, it may also include user terminal 105; Among them, the text extraction model 102 and the large language model 104 can run on the server 101, or can run on other servers. The two can run on the same server or on different servers. The server 101 can be a single server, or a server cluster composed of several servers, or one or more cloud servers in a cloud computing platform; the text extraction model 102 is used to receive the image to be evaluated sent by the server 101, and perform text extraction on the image to be evaluated to obtain image text; the large language model 104 is used to receive the image to be evaluated and the prompt text sent by the server 101, and perform authenticity verification based on the image to be evaluated and the prompt text to obtain an authenticity verification result; The credit assessment platform 103 is used to perform credit assessment based on the image text and the authenticity verification result synchronized by the server 101; An application program is installed in the user terminal 105. The user can upload the image to be evaluated through the application program installed in the user terminal 105. The user terminal 105 uploads the image to be evaluated to the server 101. The user terminal 105 can specifically be a mobile phone, a personal computer, a tablet computer, an e-book reader, a device for information interaction based on VR (Virtual Reality, virtual reality technology), a vehicle-mounted terminal, an IoT device, a wearable intelligent device, a laptop computer, a desktop computer, and so on; In this implementation environment, during the process of credit assessment processing, server 101 receives the image to be evaluated sent by user terminal 105, inputs the image to be evaluated into text extraction model 102 for text extraction to obtain image text. Server 101 performs identity verification based on the user identity information contained in the image text, and conducts content verification on the entity information contained in the image text. If the verification passes, server 101 inputs the image to be evaluated and the prompt text into large language model 104 for authenticity verification. After obtaining the authenticity verification result, it synchronizes the image text and the authenticity verification result to credit assessment platform 103 for efficient and accurate credit assessment.

[0010] One or more embodiments of a credit assessment processing method provided in this specification are as follows: Refer to Figure 2 , the credit assessment processing method provided in this embodiment, the method specifically includes steps S202 to S208.

[0011] Step S202, input the image to be evaluated into a text extraction model for text extraction to obtain structured text.

[0012] The image to be evaluated in this embodiment refers to a voucher image related to the user's credit assessment. The image to be evaluated can be an image obtained by photographing a paper voucher, such as a graduation certificate image, a real estate certificate image, a provident fund statement image, an individual income tax details image, etc. The text extraction model can be a large language model specifically trained in advance for text extraction, or other models with text extraction functions.

[0013] The structured text refers to text arranged in a structured form. When the text extraction model outputs structured text, it can output the text in a structured form, such as in the form of graduation certificate (name, graduation certificate number, school, academic degree, graduation time), real estate certificate (real estate owner's name, real estate certificate number, real estate area), provident fund (provident fund name, provident fund statement data), individual income tax (taxpayer's name, tax details).

[0014] Specifically in implementation, the image to be evaluated can be sent from the user terminal to the server. Correspondingly, after the server receives the image to be evaluated sent by the user terminal, it inputs the image to be evaluated into the text extraction model for text extraction to obtain structured text. In an optional implementation manner provided in this embodiment, the image to be evaluated is obtained in the following manner: Obtain the image to be evaluated sent by the user terminal.

[0015] Optionally, the image to be evaluated is uploaded by the user to the user terminal through an application installed on the user terminal.

[0016] Specifically, the user uploads the image to be evaluated through the credit evaluation interface provided by the application installed on the user terminal. After receiving the image to be evaluated, the user terminal sends the image to be evaluated to the server. Correspondingly, the server obtains the image to be evaluated sent by the user terminal and inputs the image to be evaluated into the text extraction model to obtain the structured text.

[0017] It should be noted that the user needs to log in to the application before uploading the image to be evaluated, that is, the upload permission of the image to be evaluated is opened after the user logs in to the application on the user terminal.

[0018] In the specific execution process, after obtaining the image to be evaluated, text extraction is performed. In order to improve the efficiency and accuracy of text extraction, the pre-trained text extraction model can be used for text recognition and text extraction. In an optional implementation provided in this embodiment, the text extraction model performs text extraction in the following manner: Perform text recognition on the image to be evaluated to obtain the text recognition result; Determine the text extraction strategy according to the text recognition result, and extract the text according to the text extraction strategy for the text recognition result to obtain the image text.

[0019] The text extraction strategy can be a structured extraction strategy. According to the text extraction strategy, the text recognition result can be extracted in a structured form. It should be noted that different types of images to be evaluated correspond to different text extraction strategies. For example, after performing text recognition on a graduation certificate image to obtain the first text recognition result, determine the first text extraction strategy according to the first text recognition result corresponding to the graduation certificate image, and extract the text recognition result of the graduation certificate image according to the first text extraction strategy. Another example is that after performing text recognition on a real estate certificate image to obtain the second text recognition result, determine the second text extraction strategy according to the second text recognition result corresponding to the real estate certificate image, and extract the second text recognition result of the real estate certificate image according to the second text extraction strategy.

[0020] Specifically, in the process of performing text recognition on the image to be evaluated, the image to be evaluated can be preprocessed first to optimize the image quality. Then, the text area in the preprocessed image is detected, and the detected text area is converted into the text recognition result. Since there may be various types of images to be evaluated, such as graduation certificate images, real estate certificate images, etc., in the process of text extraction, the text extraction strategy can be determined according to the text recognition result, and the text recognition result is extracted according to the text extraction strategy corresponding to the text recognition result to obtain the structured text.

[0021] For example, perform text recognition on the graduation certificate image to obtain text recognition results such as those of the graduation certificate (name, graduation certificate number, school, educational background, graduation time, major name, date of birth, signature). Among them, according to the text extraction strategy corresponding to the graduation certificate image, the major name, date of birth, and signature in the graduation certificate image may not need to be extracted, and only the text corresponding to the name, graduation certificate number, school, educational background, and graduation time is extracted.

[0022] Here, by leveraging the natural language processing capabilities of the text extraction model, the context relationships in the structured text can be deeply understood, and semantic changes can be captured, thereby improving the accuracy of text extraction during the text extraction process and enhancing the processing efficiency.

[0023] Step S204: Verify the identity of the user identity information included in the structured text, and verify the similarity between the entity fields included in the structured text and the preset entity fields.

[0024] After the above-mentioned step of inputting the image to be evaluated into the text extraction model for text extraction to obtain the structured text, in this step, verify the identity of the user identity information included in the structured text to detect whether the person to whom the image to be evaluated belongs is the user who uploaded the image to be evaluated, and verify the similarity between the entity fields included in the structured text and the preset entity fields to detect whether the image to be evaluated is similar to the historical evaluation image. If they are similar, it indicates that the image to be evaluated has been used for evaluation and cannot be used for credit evaluation processing again.

[0025] It should be noted that the above step S204 of verifying the identity of the user identity information included in the structured text and verifying the similarity between the entity fields included in the structured text and the preset entity fields can also be replaced by: verifying the identity of the user identity information included in the structured text and performing field verification according to the entity fields included in the structured text.

[0026] The similarity verification refers to verifying the similarity of fields. The similarity verification can be field similarity verification or field verification. Specifically, the similarity verification result can be determined by calculating the field similarity between the entity fields in the structured text and the preset entity fields.

[0027] The entity fields refer to the specific contents included in the structured text. For example, the name, graduation certificate number, school, educational background, and / or graduation time in the graduation certificate can all be used as entity fields. In the specific implementation process, in order to improve the credibility of the credit assessment process based on the image to be evaluated, after extracting the text from the image to be evaluated to obtain the structured text, the user identity information and entity fields included in the image text are verified. Specifically, identity verification is performed based on the user identity information and similarity verification is performed based on the entity fields. When both verifications pass, it is determined that the verification passes. If any one of the two verifications fails or both verifications fail, it is determined that the verification fails. The following specifically describes the process of identity verification and the process of similarity verification: (1) Identity verification In the specific process of identity verification, the user identity information included in the structured text is compared with the logged-in user information to determine whether the person to whom the image to be evaluated belongs is the same as the user who uploaded the image to be evaluated. In an optional implementation provided in this embodiment, the following method is used to verify the user identity information included in the structured text: Query the logged-in user information of the application that collected the image to be evaluated; Compare the user identity information with the logged-in user information. If they are the same, it is determined that the identity verification passes; if they are different, it is determined that the identity verification fails.

[0028] Specifically, the user who uploaded the image to be evaluated is determined by querying the logged-in user information of the application. The user identity information is compared with the logged-in user information. If they are the same, it indicates that the person to whom the image to be evaluated belongs is the same as the user who uploaded the image to be evaluated, and it is determined that the identity verification passes. If they are different, it indicates that the person to whom the image to be evaluated belongs is not the same as the user who uploaded the image to be evaluated, and it is determined that the identity verification fails.

[0029] (2) Similarity verification In the specific process of similarity verification, the field similarity between each entity field in the structured text and the preset entity field is calculated, and the text similarity is calculated based on the field weight and field similarity of each entity field. If the text similarity is less than the similarity threshold, it is determined that the similarity verification passes. If the text similarity is greater than or equal to the similarity threshold, it is determined that the similarity verification fails. In an optional implementation provided in this embodiment, the following method is used to perform similarity verification based on the entity fields included in the structured text: Calculate the field similarity between each entity field in the structured text and the preset entity field; Calculate the text similarity based on the field weight and field similarity of each entity field; If the text similarity is less than the similarity threshold, it is determined that the similarity verification has passed.

[0030] Among them, the preset entity fields can be extracted from historical structured texts; the historical structured texts can be obtained by text extraction from historical evaluation images uploaded by other users in the application and that have already undergone credit evaluation processing, where the historical evaluation images can be stored in the database of the application.

[0031] Specifically, in the process of performing similarity verification, calculate the field similarity between each entity field in the structured text and the preset entity fields, calculate the text similarity based on the field weights and field similarities of each entity field. If the text similarity is greater than or equal to the similarity threshold, it indicates that the image to be evaluated has been used for credit evaluation processing in the past and cannot be reused, and it is determined that the similarity verification has not passed. If the text similarity is less than the similarity threshold, it indicates that the image to be evaluated has not been used for credit evaluation processing in the past and can be subject to credit evaluation, and it is determined that the similarity verification has passed.

[0032] For example, if the image to be evaluated is a graduation certificate image, structured text such as graduation certificate number, school, educational attainment, and graduation time is obtained by text extraction from the graduation certificate image. Among them, each entity field is the specific content of the graduation certificate number, school, educational attainment, and graduation time. Calculate the field similarity between each entity field and the preset entity fields, calculate the text similarity based on the field weights and field similarities of each entity field. If the text similarity is less than the similarity threshold, it is determined that the similarity verification has passed.

[0033] In addition, in the process of performing similarity verification, it is also possible not to calculate the text similarity and directly calculate the field similarity between each entity field in the obtained structured text and the preset entity fields. If the field similarity is less than the similarity threshold, it is determined that the similarity verification has passed.

[0034] It should be noted that the processes of verifying the user identity information included in the structured text and performing similarity verification based on the entity fields included in the structured text can be executed in parallel. For example, after obtaining the structured text, simultaneously verify the user identity information included in the structured text and perform similarity verification based on the entity fields included in the structured text. In addition, it is also possible to perform another verification after one verification passes. For example, after the verification of the user identity information included in the structured text passes, perform similarity verification based on the entity fields included in the structured text. Another example is that after the similarity verification based on the entity fields included in the structured text passes, verify the user identity information included in the structured text. This embodiment does not make any limitations here.

[0035] Step S206, if the verification passes, input the image to be evaluated and the prompt text into the large language model for authenticity verification, and obtain the authenticity verification result.

[0036] After the above identity verification of the user identity information included in the structured text and the similarity verification based on the entity fields included in the structured text, in this step, if both verifications pass, input the image to be evaluated and the prompt text into the large language model for authenticity verification, and obtain the authenticity verification result.

[0037] The authenticity verification process refers to verifying the authenticity of the image to be evaluated, or it can also be image enhancement verification or image modification verification of the image to be evaluated to check whether the image to be evaluated is a forged image, or to check whether the image to be evaluated is an artificially processed image. For example, checking whether the image has been processed by PS (Adobe Photoshop).

[0038] Optionally, the authenticity verification process includes: extracting image features of each authenticity verification dimension and feature fusion, and determining the authenticity verification result by calculating the authenticity verification score for the fused features; specifically, first extract the image features of each authenticity verification dimension of the image to be evaluated to obtain image features, then perform fusion processing on the image features to obtain fused features, calculate the authenticity verification score based on the fused features to obtain the authenticity verification score, and finally determine the authenticity verification result according to the authenticity verification score.

[0039] During the process of specifically performing authenticity verification on the image to be evaluated, in order to improve the efficiency and accuracy of image authenticity verification, the large language model can be used to perform authenticity verification on the image to be evaluated. Further, in order to improve the authenticity verification effect of the large language model, the prompt text and the image to be evaluated can be input into the large language model together to guide the large language model to perform authenticity verification; the prompt text refers to the text used to instruct the large language model to perform a specific task or generate a specific type of answer or output. Here, the prompt text can be used to instruct the large language model to perform the authenticity verification task and generate the authenticity verification result in a preset format.

[0040] In an optional implementation manner provided in this embodiment, the following method is used for authenticity verification: Extract the image features of each authenticity verification dimension of the image to be evaluated; Perform feature fusion processing on the image features to obtain fused features; Calculate the authenticity verification score based on the fused features, and determine the authenticity verification result according to the calculated authenticity verification score.

[0041] Optionally, the authenticity verification result includes a passed authenticity verification result and a failed authenticity verification result.

[0042] During the specific execution process, the verification dimension refers to the image feature dimension that affects the authenticity of the image, and the verification dimension includes at least one of the following: image noise dimension, pixel dimension, and image structure dimension.

[0043] On this basis, during the verification process of the large language model, the image to be evaluated can be input into the feature extraction module of the large language model to extract the image features of each verification dimension. Based on the feature fusion module, the image features are subjected to feature fusion processing to obtain the fusion features. The fusion features are input into the activation function to obtain the verification score, and the preset verification result corresponding to the verification score interval where the verification score is located is used as the verification result.

[0044] For example, after performing feature fusion on the noise features, pixel features, and / or image structure features of the image to be evaluated to obtain the fusion features, calculate the verification score of the fusion features, and use the preset verification result corresponding to the verification score interval where the verification score is located as the verification result.

[0045] Here, by continuously contacting and learning new samples, the large language model can update its knowledge base in real time, adapt to and identify newly emerging material types and image anomaly types. The large language model can identify anomalies in the image by learning a large amount of data and automatically mark the content in the image that may be tampered with. In addition, combined with multimodal data processing capabilities such as image recognition, the large language model can also assist in analyzing non-text information (such as images, tables, etc.), further enriching the verification processing dimension and improving the overall security. In addition, using transfer learning and incremental learning techniques, the large language model can quickly adapt to new requirements and technical backgrounds while maintaining the existing knowledge, making the entire system more flexible and scalable.

[0046] Step S208, synchronize the structured text and the verification result to the credit assessment platform for credit assessment processing.

[0047] After obtaining the verification result by inputting the image to be evaluated and the prompt text into the large language model for verification processing as described above, in this step, the structured text and the verification result are synchronized to the credit assessment platform for credit assessment processing by the credit assessment platform. After that, the credit assessment platform may return a credit assessment failure result or a credit assessment result.

[0048] During the specific execution process, the structured text and the verification result can be synchronized to the credit assessment platform by means of interface calls, so that the credit assessment platform can perform credit assessment processing. In an optional implementation manner provided in this embodiment, the credit assessment platform performs credit assessment processing in the following manner: If the verification result is a successful verification, extract the entity fields in the structured text for credit assessment; calculate the credit limit variable according to the preset assessment rules and the entity fields, and generate a credit assessment result including the credit limit variable; If the verification result is a failed verification, return a credit assessment failure result.

[0049] Specifically, in the case where the verification result is a successful verification, the entity fields in the structured text for credit assessment can be extracted, and the credit limit variable corresponding to the entity fields can be found from the preset assessment rules, and a credit assessment result including the credit limit variable can be generated.

[0050] For example, when the credit assessment platform identifies that the verification result is a successful verification, it extracts the first education information in the structured text, and finds from the preset assessment rules that the credit limit variable corresponding to the first education information is x. Therefore, when the education information in the extracted structured text is the first education information, the credit limit variable is determined to be x.

[0051] In addition, during the process of the credit assessment platform performing credit assessment processing, a credit score can also be generated, and the credit limit variable can be calculated according to the credit score. In another alternative implementation provided in this embodiment, the credit assessment platform performs credit assessment processing in the following manner: If the verification result is a successful verification, extract the entity fields in the structured text for credit assessment; calculate the credit score according to the scoring rules and the extracted entity fields, and generate a credit assessment result including the credit limit variable based on the credit score; If the verification result is a failed verification, return a credit assessment failure result.

[0052] Specifically, when the credit assessment platform identifies that the verification result is a successful verification, it performs credit assessment processing. During the specific process of performing credit assessment processing, it extracts the entity fields in the structured text for credit assessment, calculates the credit score according to the scoring rules and the extracted entity fields, and generates a credit assessment result including the credit limit variable based on the credit score.

[0053] During the specific execution process, after the credit assessment platform generates the credit limit variable, it generates a credit assessment result including the credit limit variable and returns it to the server. Correspondingly, the server receives the credit assessment result including the credit limit variable returned by the credit assessment platform, and updates the remaining credit limit of the user. In an alternative implementation provided in this embodiment, the remaining credit limit is updated in the following manner: Calculate the current credit limit of the user according to the available credit limit of the user and the credit assessment result returned by the credit assessment platform; Update the available credit limit based on the current credit limit to obtain an updated credit limit.

[0054] Specifically, receive the credit assessment result containing the credit limit variable returned by the credit assessment platform, and query the available credit limit of the user. Calculate the current credit limit based on the credit limit variable and the available credit limit in the credit assessment result, and update the available credit limit based on the current credit limit to obtain the updated credit limit.

[0055] For example, if the credit limit variable in the received credit assessment result is n, and the available credit limit of the queried user is x, then take x + n as the current credit limit, and update the remaining credit limit to x + n.

[0056] Furthermore, in order to enhance the user's perception, after updating the user's remaining credit limit, an amount update message containing the updated credit limit can be generated and sent to the user terminal. In an optional implementation provided in this embodiment, after the step of updating the available credit limit based on the current credit limit, it further includes: Generate an amount update reminder containing the updated credit limit; Send the amount update reminder to the user terminal to display the amount update message through the application installed on the user terminal.

[0057] In addition, when the credit assessment platform identifies that the verification result is a failed verification, the credit assessment platform can also return a credit assessment failure result to the server. Correspondingly, the server receives the credit assessment failure result returned by the credit assessment platform and performs a manual review on the image to be evaluated to avoid misjudgment and improve reliability. In an optional implementation provided in this embodiment, after the step of synchronizing the structured text and the verification result to the credit assessment platform for credit assessment, it further includes: If the credit assessment fails, mark the image to be evaluated and send the obtained marked image to the reviewer for the reviewer to perform a review process on the marked image.

[0058] Specifically, if a credit assessment failure result returned by the credit assessment platform is received, mark the image to be evaluated and send the obtained marked image to the reviewer for the reviewer to perform a manual review on the image to be evaluated; in addition, after receiving the assessment failure result returned by the credit assessment platform, the image to be evaluated may not be marked and directly sent to the reviewer for a secondary review.

[0059] It should be noted that through large-scale data training, the large language model can also explore potential trends and patterns in the process of verifying the images to be evaluated, providing important reference for subsequent risk warnings and decision support. For example, the large language model records and reports the types of image anomalies that frequently appear during the verification process, and the auditors take corresponding preventive measures for the types of image anomalies that frequently appear; among them, the image anomaly type can be anomalies such as the name being altered in the graduation certificate image.

[0060] It should also be noted that the above-mentioned step S208 of synchronizing the structured text and the verification result to the credit assessment platform for credit assessment processing can also be replaced by: synchronizing the structured text to the credit assessment platform for credit assessment processing when the verification is passed, that is, after obtaining the verification result, the server identifies the verification result, and if the identified verification result is that the verification is passed, the structured text is synchronized to the credit assessment platform for credit assessment processing.

[0061] In summary, the present embodiment provides a credit assessment processing method, which first inputs the image to be evaluated into a text extraction model for text extraction, so as to extract relatively accurate structured text with the help of the text extraction capability of the text extraction model, and then verifies the identity of the user identity information contained in the structured text, and verifies the similarity based on the entity fields contained in the structured text, so as to ensure the security of the credit assessment. If the verification is passed, the image to be evaluated and the prompt text are input into a large language model for verification processing, so as to check the authenticity of the image to be evaluated and ensure that the image to be evaluated has not been abnormally modified. After obtaining the verification result, the structured text and the verification result are synchronized to the credit assessment platform, so as to perform credit assessment processing through the credit assessment platform, thereby improving the automation and accuracy of the credit assessment, and realizing efficient evaluation and flexible update of the credit limit.

[0062] The following uses a credit assessment processing method provided in this embodiment as an example in a credit limit assessment processing scenario. Figure 3 , for further explanation of the credit assessment processing method provided in this embodiment, see Figure 3 , a credit assessment processing method applied to a credit limit assessment processing scenario specifically includes the following steps.

[0063] Step S302: input the image to be evaluated into a text extraction model to extract text and obtain structured text.

[0064] Step S304: querying the login user information of the application program that collects the image to be evaluated.

[0065] Step S306, comparing the user identity information contained in the structured text with the login user information, and if they are consistent, determining that the identity verification is successful.

[0066] Step S308: Calculate the field similarity between each entity field in the structured text and the preset entity field.

[0067] Step S310: If the field similarity is less than the similarity threshold, determine that the similarity verification passes.

[0068] Step S312: Extract the image features of each verification dimension of the image to be evaluated, and perform feature fusion processing to obtain the fusion features.

[0069] Step S314: Calculate the verification score based on the fusion features, and determine the verification result according to the verification score.

[0070] Step S316: In the case of successful verification, synchronize the structured text to the credit assessment platform for credit assessment processing.

[0071] Step S318: Calculate the current credit limit of the user according to the available credit limit of the user and the credit assessment result returned by the credit assessment platform.

[0072] Step S320: Update the available credit limit based on the current credit limit to obtain the updated credit limit.

[0073] Step S322: Generate a limit update reminder including the updated credit limit.

[0074] Step S324: Send the limit update reminder to the user terminal to display the limit update information through the application installed on the user terminal.

[0075] It should be noted that any one step or any combination of multiple steps among steps S302 to S324 can be combined with any one step or any combination of multiple steps of the above steps S202 to S208 according to the needs of implementation and deployment; in addition, according to the actual deployment needs, any one or any combination of technical features among steps S302 to S324 can be combined with any one or more technical features provided by the above steps S202 to S208 to form a new implementation; or, any one or any combination of technical features among steps S302 to S324 can also be replaced by any one or more technical features provided by the above steps S202 to S208 according to the actual deployment needs to form a new implementation, which will not be elaborated here one by one.

[0076] An embodiment of a credit assessment processing device provided in this specification is as follows: In the above embodiment, a credit assessment processing method is provided. Correspondingly, a credit assessment processing device is also provided, which will be described below with reference to the accompanying drawings.

[0077] Reference Figure 4 , which shows a schematic diagram of an embodiment of a credit evaluation processing device provided in this embodiment.

[0078] Since the device embodiment corresponds to the method embodiment, the description is relatively simple. For the relevant parts, please refer to the corresponding description of the method embodiment provided above. The device embodiments described below are merely illustrative.

[0079] This embodiment provides a credit evaluation processing device, which includes: A text extraction module 402, configured to input the image to be evaluated into a text extraction model for text extraction to obtain structured text; A verification module 404, configured to verify the user identity information included in the structured text, and verify the similarity between the entity fields included in the structured text and the preset entity fields; If the verification is passed, the authenticity verification processing module 406 is run. The authenticity verification processing module 406 is configured to input the image to be evaluated and the prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: extracting and fusing image features for each authenticity verification dimension, and determining the authenticity verification result by calculating the authenticity verification score for the fused features; A credit evaluation module 408, configured to synchronize the structured text and the authenticity verification result to a credit evaluation platform for credit evaluation processing.

[0080] An embodiment of a credit evaluation processing device provided in this specification is as follows: Corresponding to the above-described credit evaluation processing method, based on the same technical concept, one or more embodiments of this specification also provide a credit evaluation processing device, which is used to execute the above-provided credit evaluation processing method. Figure 5 It is a schematic structural diagram of a credit evaluation processing device provided by one or more embodiments of this specification.

[0081] A credit evaluation processing device provided in this embodiment includes: Such as Figure 5As shown, credit assessment processing devices can vary significantly in configuration or performance. They can include one or more processors 501 and a memory 502. The memory 502 can store one or more stored application programs or data. Among them, the memory 502 can be short-term storage or persistent storage. The application programs stored in the memory 502 can include one or more modules (not shown in the figure), and each module can include a series of computer-executable instructions in the credit assessment processing device. Further, the processor 501 can be set to communicate with the memory 502 and execute a series of computer-executable instructions in the memory 502 on the credit assessment processing device. The credit assessment processing device can also include one or more power supplies 503, one or more wired or wireless network interfaces 504, one or more input / output interfaces 505, one or more keyboards 506, etc.

[0082] In a specific embodiment, the credit assessment processing device includes a memory and one or more programs. One or more of the programs are stored in the memory, and one or more of the programs can include one or more modules. Each module can include a series of computer-executable instructions in the credit assessment processing device and is configured to be executed by one or more processors. The one or more programs include the following computer-executable instructions for: Input the image to be evaluated into a text extraction model for text extraction to obtain structured text; Verify the user identity information included in the structured text and verify the similarity between the entity fields included in the structured text and the preset entity fields; If the verification passes, input the image to be evaluated and the prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: extracting and fusing image features in each authenticity verification dimension, and determining the authenticity verification result by calculating the authenticity verification score for the fused features; Synchronize the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing.

[0083] An embodiment of a computer-readable storage medium provided in this specification is as follows: Corresponding to the above-described credit assessment processing method, based on the same technical concept, one or more embodiments of this specification also provide a computer-readable storage medium.

[0084] The computer-readable storage medium provided in this embodiment is used to store computer-executable instructions, and the computer-executable instructions, when executed, implement the following process: Input the image to be evaluated into a text extraction model for text extraction to obtain structured text; Verify the user identity information contained in the structured text, and verify the similarity between the entity fields contained in the structured text and the preset entity fields; If the verification passes, input the image to be evaluated and the prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: extracting and fusing image features in each authenticity verification dimension, and determining the authenticity verification result by calculating the authenticity verification score for the fused features; Synchronize the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing.

[0085] It should be noted that the embodiments of a computer-readable storage medium in this specification and the embodiments of a credit assessment processing method in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the corresponding method described above, and the repeated parts will not be elaborated.

[0086] An embodiment of a computer program product provided in this specification is as follows: Corresponding to the above-described credit assessment processing method, based on the same technical concept, one or more embodiments of this specification also provide a computer program product.

[0087] A computer program product includes a computer program / instructions, and when the computer program / instructions are executed by a processor, the following steps are implemented: Input the image to be evaluated into a text extraction model for text extraction to obtain structured text; Verify the user identity information contained in the structured text, and verify the similarity between the entity fields contained in the structured text and the preset entity fields; If the verification passes, input the image to be evaluated and the prompt text into a large language model for authenticity verification processing to obtain an authenticity verification result; the authenticity verification processing includes: extracting and fusing image features in each authenticity verification dimension, and determining the authenticity verification result by calculating the authenticity verification score for the fused features; Synchronize the structured text and the authenticity verification result to a credit assessment platform for credit assessment processing.

[0088] It should be noted that the embodiments of a computer program product in this specification and the embodiments of a credit assessment processing method in this specification are based on the same inventive concept. Therefore, the specific implementation of this embodiment can refer to the implementation of the corresponding method described above, and the repeated parts will not be elaborated.

[0089] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments. For example, the device embodiment, the equipment embodiment, the computer-readable storage medium embodiment, and the computer program product embodiment are all similar to the method embodiment, so the description is relatively simple. To read the relevant content in the device embodiment, the equipment embodiment, the computer-readable storage medium embodiment, and the computer program product embodiment, please refer to the corresponding part of the method embodiment for explanation.

[0090] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] In the 1930s, it was obvious to distinguish whether an improvement to a technology was a hardware improvement (e.g., improvement to circuit structures such as diodes, transistors, switches, etc.) or a software improvement (improvement to method processes). However, with the development of technology, many improvements to method processes today can be regarded as direct improvements to hardware circuit structures. Almost all designers obtain the corresponding hardware circuit structures by programming the improved method processes into the hardware circuits. Therefore, it cannot be said that an improvement to a method process cannot be implemented with a hardware entity module. For example, a Programmable Logic Device (PLD) (e.g., a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can program themselves to "integrate" a digital system onto a single PLD, without having to ask a chip manufacturer to design and fabricate a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compilers used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a Hardware Description Language (HDL). There is not only one type of HDL, but many types, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones currently are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also be aware that by simply performing a little logical programming on the method process using the above-mentioned several hardware description languages and programming it into an integrated circuit, it is easy to obtain the hardware circuit that implements the logical method process.

[0092] The controller can be implemented in any suitable manner. For example, the controller can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of the controller include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that in addition to implementing the controller in the form of pure computer-readable program code, it is entirely possible to make the controller implement the same function in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or structures within the hardware component.

[0093] The systems, devices, modules, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0094] For the convenience of description, when describing the above devices, they are described separately as various units according to their functions. Of course, when implementing the embodiments of this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0095] Those skilled in the art should understand that one or more embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, one or more embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this specification can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code.

[0096] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the specification. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0097] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0098] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 flow or multiple flows and / or blocks Figure 1 block or multiple blocks.

[0099] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0100] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash RAM. The memory is an example of computer-readable media.

[0101] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include transitory media such as modulated data signals and carrier waves.

[0102] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of further restrictions, the elements defined by the sentence "includes at least one ..." do not exclude the presence of other identical elements in the process, method, commodity or device including the elements.

[0103] One or more embodiments of the present specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. One or more embodiments of the present specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0104] The above description is only an embodiment of this document and is not intended to limit this document. For those skilled in the art, this document may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this document should be included in the scope of the claims of this document.

Claims

1. A credit assessment processing method, characterized in that: The method comprises: Input the image to be evaluated into the text extraction model to extract text and obtain structured text; Performing identity verification on the user identity information contained in the structured text, and performing similarity verification on the entity field contained in the structured text and the preset entity field; If the verification is passed, the image to be evaluated and the prompt text are input into the large language model for verification processing to obtain a verification result; the verification processing includes: extracting image features and fusion of features in each verification dimension, and determining the verification result by calculating the verification score of the fusion feature; The structured text and the verification result are synchronized to a credit evaluation platform for credit evaluation processing.

2. The credit assessment processing method according to claim 1, characterized in that: The identity verification of the user identity information contained in the structured text includes: Querying the login user information of the application program that collected the image to be evaluated; The user identity information is compared with the login user information, and if they are consistent, it is determined that the identity verification has passed.

3. The credit assessment processing method according to claim 1, characterized in that: The similarity verification between the entity field included in the structured text and the preset entity field includes: Calculating the field similarity between each entity field in the structured text and a preset entity field; Calculating text similarity based on the field weights of the entity fields and the field similarity; If the text similarity is less than the similarity threshold, it is determined that the similarity verification has passed.

4. The credit assessment processing method according to claim 1, characterized in that: The verification process is implemented in the following way: Extracting image features of each verification dimension of the image to be evaluated; Performing feature fusion processing on the image features to obtain the fused features; A verification score is calculated based on the fusion feature, and the verification result is determined according to the calculated verification score.

5. The credit assessment processing method according to claim 1, characterized in that: The credit assessment process includes: If the verification result is that the verification is passed, extracting the entity field used for credit evaluation in the structured text; The credit limit variable is calculated according to the preset evaluation rules and the entity field, and a credit evaluation result including the credit limit variable is generated.

6. The credit assessment processing method according to claim 1, characterized in that: After the step of synchronizing the structured text and the verification result to the credit evaluation platform for credit evaluation processing is executed, the method further includes: Calculate the current credit limit of the user according to the available credit limit of the user and the credit evaluation result returned by the credit evaluation platform; The available credit limit is updated based on the current credit limit to obtain an updated credit limit.

7. The credit assessment processing method according to claim 6, characterized in that: After the step of updating the available credit limit based on the current credit limit and obtaining the updated credit limit is performed, the method further includes: generating a credit limit update reminder including the updated credit limit; The credit limit update reminder is sent to a user terminal, so that the credit limit update information is displayed through an application installed on the user terminal.

8. The credit assessment processing method according to claim 1, characterized in that: After the step of synchronizing the structured text and the verification result to the credit evaluation platform for credit evaluation processing is executed, the method further includes: If the credit evaluation fails, the image to be evaluated is marked, and the marked image is sent to the auditor so that the auditor can review the marked image.

9. The credit assessment processing method according to claim 1, characterized in that: The text extraction is implemented in the following way: Performing text recognition on the image to be evaluated to obtain a text recognition result; The text recognition result is subjected to text extraction according to a text extraction strategy to obtain the structured text.

10. The credit assessment processing method according to claim 1, characterized in that: Before the step of inputting the image to be evaluated into the text extraction model to extract text and obtain structured text is performed, the step further includes: Acquire the image to be evaluated sent by the user terminal; The image to be evaluated is uploaded to the user terminal by a user through an application installed on the user terminal.

11. A credit assessment processing device, characterized in that: The device comprises: A text extraction module is configured to input the image to be evaluated into a text extraction model to extract text and obtain structured text; A verification module configured to verify the identity of the user contained in the structured text, and to verify the similarity between the entity field contained in the structured text and the preset entity field; If the verification is passed, the verification processing module is run, and the verification processing module is configured to input the image to be evaluated and the prompt text into the large language model for verification processing to obtain a verification result; the verification processing includes: performing image feature extraction and feature fusion of each verification dimension, and determining the verification result by calculating the verification score of the fused feature; The credit evaluation module is configured to synchronize the structured text and the verification result to a credit evaluation platform for credit evaluation processing.

12. A credit assessment processing device, characterized in that: The device comprises: a processor; and a memory configured to store computer-executable instructions that, when executed, cause the processor to: Input the image to be evaluated into the text extraction model to extract text and obtain structured text; Performing identity verification on the user identity information contained in the structured text, and performing similarity verification on the entity field contained in the structured text and the preset entity field; If the verification is passed, the image to be evaluated and the prompt text are input into the large language model for verification processing to obtain a verification result; the verification processing includes: extracting image features and fusion of features in each verification dimension, and determining the verification result by calculating the verification score of the fusion feature; The structured text and the verification result are synchronized to a credit evaluation platform for credit evaluation processing.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium is used to store computer-executable instructions, and the computer-executable instructions implement the steps of the method of claim 1 when executed.

Images