Bank credit liability affirmation report generation method and device

Abstract

The invention provides a bank credit responsibility affirmation report generation method and device. The method comprises the following steps: acquiring bank credit source data; obtaining customer number information and debt item number information of the credit and loan service, and obtaining an SQL query statement of the credit and loan service according to the customer number information and the debt item number information of the credit and loan service in combination with a preset cue word; determining target credit data corresponding to the customer number information and the debt item number information according to bank credit source data by using an SQL query statement of a credit service; inputting the target credit data into a bank credit business large model to obtain a bank credit responsibility affirmation report; through fusion of the large model and the cue word, a credit report generation scheme which is adaptive to credit data characteristics, balances generation efficiency and accuracy and reduces a technical use threshold is determined based on the SQL statement, and technical support is provided for automatic generation of a bank credit responsibility affirmation report.

Description

Technical Field

[0001] This invention belongs to the field of bank big data processing technology, and in particular relates to a method and apparatus for generating bank credit liability determination reports. Background Technology

[0002] In banking lending scenarios, the automated generation of credit liability determination reports is a key step in improving business efficiency and standardizing the determination process. Currently, the mainstream solution for the automated generation of such reports in the industry relies on natural language processing technology to realize the core process.

[0003] However, traditional solutions have significant limitations in practical applications: their implementation and use are highly dependent on technical personnel. Not only do professionals need to master development languages ​​such as Java, but they also need to have professional knowledge related to NLP technology to complete tasks such as data tag definition and template adaptation. This makes it difficult for business personnel to operate directly, resulting in a high barrier to entry and making it difficult to meet the needs of efficient and flexible banking operations. Summary of the Invention

[0004] This invention provides a method for generating bank credit liability determination reports. By fusing a large model with prompt words, and based on SQL statements, a credit report generation scheme is determined that adapts to the characteristics of credit data, balances generation efficiency and accuracy, and lowers the technical barrier to entry. This provides technical support for the automated generation of bank credit liability determination reports. The method includes:

[0005] Obtain bank credit source data, which includes business registration data, debt data, and personnel information;

[0006] Obtain customer ID and debt number information for credit business; based on the customer ID and debt number information for credit business, and combined with preset prompt words, obtain the SQL query statement for credit business.

[0007] Using SQL query statements for credit business, the target credit data corresponding to customer number information and debt number information is determined based on bank credit source data;

[0008] The target credit data is input into the bank's credit business model to obtain a bank credit liability determination report; the bank's credit business model is obtained by training a neural network model based on historical bank credit data.

[0009] This invention provides a bank credit liability determination report generation device. By fusing a large model with prompt words, and based on SQL statements, it determines a credit report generation scheme that adapts to the characteristics of credit data, balances generation efficiency and accuracy, and lowers the technical barrier to entry. This provides technical support for the automated generation of bank credit liability determination reports. The bank credit liability determination report generation device includes:

[0010] The data acquisition module is used to acquire bank credit source data, which includes business registration data, debt data, and personnel information.

[0011] The SQL query statement generation module is used to obtain customer ID information and debt number information for credit business, and to generate SQL query statements for credit business based on the customer ID information and debt number information and preset prompt words.

[0012] The target credit data determination module is used to determine the target credit data corresponding to customer number information and debt number information based on the bank's credit source data using SQL query statements for credit business.

[0013] The bank credit liability determination report generation module is used to input target credit data into the bank credit business big model to obtain the bank credit liability determination report; the bank credit business big model is obtained by training a neural network model based on historical bank credit data.

[0014] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for generating a bank credit liability determination report.

[0015] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for generating a bank credit liability determination report.

[0016] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-mentioned method for generating a bank credit liability determination report.

[0017] In this embodiment of the invention, bank credit source data is acquired, including business registration data, debt data, and personnel information; customer ID information and debt ID information for credit business are acquired; based on the customer ID information and debt ID information for credit business, and combined with preset prompt words, an SQL query statement for credit business is obtained; using the SQL query statement for credit business, target credit data corresponding to the customer ID information and debt ID information is determined based on the bank credit source data; the target credit data is input into a large-scale bank credit business model to obtain a bank credit liability determination report; the large-scale bank credit business model is obtained by training a neural network model based on historical bank credit data; this embodiment of the invention, through the fusion of the large-scale model and prompt words, determines a credit report generation scheme that adapts to the characteristics of credit data, balances generation efficiency and accuracy, and lowers the technical threshold for use, based on SQL statements, providing technical support for the automated generation of bank credit liability determination reports. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0019] Figure 1 This is a flowchart of the method for generating a bank credit liability determination report in an embodiment of the present invention;

[0020] Figure 2 This is a specific example diagram illustrating the determination of target credit data in an embodiment of the present invention;

[0021] Figure 3 This is a specific example diagram of obtaining a bank credit liability determination report in an embodiment of the present invention;

[0022] Figure 4 This is a structural example diagram of the bank credit liability determination report generation device in an embodiment of the present invention;

[0023] Figure 5 This is a specific example diagram of the structure of the bank credit liability determination report generation device in an embodiment of the present invention;

[0024] Figure 6 This is a specific example diagram of the structure of the bank credit liability determination report generation device in an embodiment of the present invention;

[0025] Figure 7 This is a specific example diagram of the structure of the bank credit liability determination report generation device in an embodiment of the present invention;

[0026] Figure 8 This is a structural diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0028] The acquisition, transmission, storage, use, and processing of data in the technical solution of this invention all comply with relevant regulations.

[0029] It should be noted that in the embodiments of the present invention, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of the present invention. However, they do not mean that the applicant has used or necessarily used the solution.

[0030] As mentioned earlier, in the existing technology, the automated writing of bank credit reports is mainly achieved through natural language processing (NLP) technology. This requires tagging key data items in the report, creating standardized templates, and then using SQL statements to filter data and replace tags. However, this solution has the drawbacks of requiring customized templates and lacking universality. Moreover, it has a high barrier to entry, requiring proficiency in programming languages ​​such as Java and NLP technology to operate. This results in reliance on manual sorting for credit liability determination, leading to problems such as long writing cycles, low efficiency, and high error rates.

[0031] To address this issue, the inventors discovered that the integration of large models and prompt word engineering can replace traditional NLP solutions. By integrating credit-related data through a data warehouse and linking and integrating data with big data cloud, the large models can achieve functions such as converting natural language to SQL to retrieve data, converting data to JSON structured format, and intelligently generating reports according to templates through prompt word engineering. This forms a fully automated solution that can generate liability determination reports with just one click by only inputting the customer number and debt number, reducing the threshold for use and development costs.

[0032] Figure 1 This is a flowchart of the method for generating a bank credit liability determination report in an embodiment of the present invention, such as... Figure 1 As shown, the method for generating the bank's credit liability assessment report includes:

[0033] Step 101: Obtain bank credit source data, which includes business registration data, debt data, and personnel information;

[0034] Step 102: Obtain customer ID information and debt item ID information for the credit business. Based on the customer ID information and debt item ID information for the credit business, and combined with the preset prompt words, obtain the SQL query statement for the credit business.

[0035] Step 103: Using the SQL query statement of credit business, determine the target credit data corresponding to the customer number information and debt number information based on the bank credit source data;

[0036] Step 104: Input the target credit data into the bank credit business big model to obtain the bank credit responsibility determination report; the bank credit business big model is obtained by training a neural network model based on historical bank credit data.

[0037] Depend on Figure 1 As shown in the process, in this embodiment of the invention, firstly, bank credit source data containing business registration data, debt data, and personnel information is obtained; then, customer ID and debt ID information of the credit business are obtained, and an SQL query statement is obtained by combining preset prompt words; the corresponding target credit data is filtered out from the bank credit source data using the SQL query statement; finally, the target credit data is input into the bank credit business big model trained with historical bank credit data to generate a bank credit liability determination report.

[0038] Compared to existing technologies that use NLP to tag key data items in reports, create standardized templates, and then use SQL statements to filter data and replace tags, this new approach combines prompt word engineering to convert natural language into SQL for data querying. It then generates reports based on a trained large-scale model of bank credit business. This eliminates the need for customized template development and requires no knowledge of Java or NLP technologies. It has a low barrier to entry, high reusability, and can automate the entire process of generating bank credit liability determination reports. This solves the problems of long writing cycles, low efficiency, and high error rates caused by relying on manual review for credit liability determination.

[0039] In step 101, bank credit source data is obtained, which includes business registration data, debt data, and personnel information.

[0040] In a specific embodiment, bank credit source data containing business registration data, debt data, and personnel information is obtained, including:

[0041] Data source confirmed:

[0042] Relying on the head office-level big data cloud platform, in which the big data cloud platform plays the role of data bus, integrating the bank's internal data warehouse and isolating it from other systems; as a data foundation, it extracts credit-related source data from the data warehouse;

[0043] The branch-level data platform integrates target data from the big data cloud, performs preliminary processing on the extracted source data, and forms data support for the generation of the responsibility determination report.

[0044] Specific data acquisition:

[0045] Business registration data: obtained from enterprise business registration data tables integrated by big data cloud, including basic business registration information such as enterprise name, unified social credit code, registered address, establishment time, business scope, and equity structure;

[0046] Debt data: Extracted from the bank's internal debt data table, covering core debt information for credit business such as loan amount, loan term, interest rate, loan disbursement date, repayment method, guarantee type, loan purpose, and overdue status;

[0047] Personnel Information: Collect data on personnel involved in the entire credit business process, including the names and positions of approvers, approval opinions, information of handlers, identity information of borrowers and guarantors, and other relevant personnel information;

[0048] Supplementary data: Simultaneously acquire credit inquiry records and other data related to credit liability determination. Together with the above three types of data, they constitute complete bank credit source data, providing comprehensive data support for subsequent target credit data inquiries.

[0049] In this embodiment, after obtaining bank credit source data, the method further includes:

[0050] The bank credit source data is cleaned, deduplicated, matched, and standardized in terms of fields.

[0051] In a specific implementation, data cleaning involves: relying on the branch-level data platform to remove invalid data from business registration data, debt data, personnel information, and credit inquiry records, such as incorrectly formatted numbers, records missing key information, and logically contradictory data; correcting data entry errors, such as typos and numerical deviations, to ensure the accuracy and completeness of the source data and lay the foundation for subsequent data use.

[0052] Data deduplication: Using customer ID and debt ID as core key fields, duplicate credit source data is identified and removed. For example, duplicate business registration information records of the same enterprise and duplicate debt data entries of the same debt are deleted to avoid redundant data affecting query efficiency and report generation accuracy.

[0053] Association and Matching: By linking data from different data tables through customer ID, debt ID, and other related items, we can achieve the binding of business registration data with corresponding debt data, the correspondence between debt data and information of personnel involved in the entire process, and the matching of debt data with credit inquiry records. This forms an association data table with customers and debts as the core, which meets the needs of liability determination reports for multi-dimensional data association queries.

[0054] Field standardization: According to the preset data standards, the field format and naming conventions of various data types are unified. For example, date data in different formats are unified into the "YYYY-MM-DD" format, and the unit for monetary fields is unified into "yuan". The description of classification fields such as personnel position and guarantee type is standardized to ensure that the data output of the data platform is in a unified format, which is convenient for subsequent SQL queries, JSON format conversion and report content extraction and filling in the large model.

[0055] In step 102, the customer ID information and debt item ID information of the credit business are obtained. Based on the customer ID information and debt item ID information of the credit business, combined with the preset prompt words, the SQL query statement of the credit business is obtained.

[0056] In a specific embodiment, a credit liability determination request is initiated through the branch-level business system. The customer number and debt number of the target credit business are entered into the system interaction interface. After receiving and verifying the format validity of the two types of numbers, such as if they conform to the internal numbering rules, they are submitted to the large model as core query conditions.

[0057] Preset prompt design: The preset prompts adopt a structure of "background + role + input description + task breakdown + output constraints + sample + constraints".

[0058] Background and Role: "You are a database operations expert, currently performing a bank credit data query."

[0059] Input description: Specify the contents and data dictionary of data tables such as the business registration table, debt table, and personnel information table in the database. For example, the business registration table contains fields such as company name and unified social credit code, and the debt table contains fields such as loan amount and overdue status.

[0060] Task breakdown: Only SQL statements for data querying need to be written, following the SELECT query syntax rules, using the input customer number and debt number as the association conditions to query the target fields in the relevant data tables;

[0061] Output constraints: Only output complete, directly executable SQL query statements without any additional explanation;

[0062] Example: A sample query is provided: "SELECT Company Name, Loan Amount FROM Business Table JOIN Debt Table ON Business Table. Customer Number = Debt Table. Customer Number WHERE Business Table. Customer Number = 'AAA' AND Debt Table. Debt Number = 'YYY'".

[0063] Constraints: Strictly follow the sample format for output; no free interpretation or addition of extra statements is allowed.

[0064] SQL statement generation: After receiving customer ID, debt number, and the aforementioned preset prompt words from the business system, the large model automatically translates the query requirements described in natural language into a standardized SQL query statement based on the logic and syntax rules constrained by the prompt words. This statement uses customer ID and debt number as the core filtering conditions and can query corresponding credit data in related data tables such as the business registration table, debt table, and personnel information table.

[0065] Figure 2 This is a specific example diagram illustrating the determination of target credit data in an embodiment of the present invention, such as... Figure 2 As shown, using SQL queries for credit operations, the target credit data corresponding to customer ID information and debt ID information can be determined based on bank credit source data. This can include:

[0066] Step 201: Using the SQL query statement of credit business, match and extract the corresponding industrial and commercial data, debt data and personnel information from the bank credit source data according to the customer number information and debt number information;

[0067] Step 202: Perform data consistency verification on the matched business registration data, debt data, and personnel information. If the verification is successful, the data is confirmed as the target credit data.

[0068] In a specific embodiment, relevant data is matched and extracted:

[0069] The SQL query connects to the Oracle database service of the branch-level data platform. Using the input customer number and debt number as the core association conditions, it accurately matches and extracts the corresponding enterprise's business registration data, such as enterprise name, unified social credit code, equity structure, etc., debt data, such as loan amount, loan term, overdue status, etc., and personnel information, such as the name of the approver, the information of the handler, and the identity information of the borrower, from the bank credit source data after it has been cleaned, deduplicated, matched, and standardized. At the same time, it also extracts the associated credit inquiry record data.

[0070] The extracted data is output in CSV format, ensuring that the data fields are consistent with the target fields specified in the SQL query statement, and covering the multi-dimensional data information required for the liability determination report.

[0071] Data consistency verification and target data determination:

[0072] Perform consistency checks on the extracted CSV format data: using customer number and debt number as the core of the check, verify whether the association logic between business registration data and debt data is reasonable. For example, whether the customer name corresponding to the debt data is consistent with the company name in the business registration data, whether the correspondence between personnel information and debt data is accurate, such as whether the approval personnel information matches the debt approval record, and whether the data field format meets the standardization requirements.

[0073] After verification, the CSV format data is converted to JSON format using the preset prompt word project, and the structured JSON data is output. This JSON data is the target credit data. If the verification finds that the data is inconsistent or missing, it is returned to the data platform to be extracted again until the data meets the requirements.

[0074] In this embodiment, after determining the target credit data corresponding to the customer number information and debt number information based on the bank's credit source data using the SQL query statement of the credit business, the method further includes:

[0075] The target credit data is converted into JSON structured data; the JSON structured data is categorized into business data, debt data, and personnel information.

[0076] In a specific embodiment, a preset JSON format conversion prompt is used: the prompt adopts a structure of "background + role + input description + task breakdown + output constraints + example".

[0077] Background and Role: "You are an Excel document analysis expert, currently performing Excel document parsing work."

[0078] Input description: The input is the target credit data (CSV format) that has passed the consistency check, including business registration data, debt data, personnel information and related credit inquiry records;

[0079] Task breakdown: Divide the CSV format data into three categories: business data, debt data, and personnel information. Extract the fields and corresponding data for each category and convert them into JSON strings in key-value pair format.

[0080] Output constraints: The JSON string should use "Business Data", "Debt Data" and "Personnel Information" as the first-level keys, the specific data items under each category as the second-level keys, and the corresponding data as the values. The structure should be clear, with no fields missing and no extra redundant information.

[0081] Example: A sample of the following is provided: "{'Business Registration Data':{'Company Name':'AAA','Unified Social Credit Code':'AAA'},'Debt Data':{'Loan Amount':'AAA'},'Personnel Information':{'Approving Personnel Name':'AAA','Handling Personnel Information':'AAA'}}".

[0082] Perform format conversion: Input the target credit data in CSV format along with the preset prompts mentioned above into the large model. Based on the prompt constraints, the large model automatically identifies the data categories and splits the fields, accurately mapping each category of data to the corresponding key-value pairs, generating JSON structured data divided into business data, debt data, and personnel information, ensuring that the data is organized in a standardized manner and is easy to extract and generate reports for the subsequent large model.

[0083] Figure 3 This is a specific example diagram of obtaining a bank credit liability determination report in an embodiment of the present invention, as shown below. Figure 3 As shown, inputting the target credit data into the bank's credit business model yields a bank credit liability determination report, which may include:

[0084] Step 301: Import a preset liability determination report template into the bank's credit business big model. The liability determination report template includes a borrower information section, a guarantee information section, and a loan balance information section.

[0085] Step 302: Input the target credit data into the bank credit business model to obtain the bank credit business model based on the preset liability determination report template. Extract the corresponding credit data of each segment from the target credit data to obtain the bank credit liability determination report.

[0086] In a specific embodiment, a preset liability determination report template is imported:

[0087] Import a pre-set standardized liability determination report template into the bank's credit business model. The template clearly divides the borrower information section, guarantee information section, and loan balance information section. Each section contains a fixed writing format and content framework. For example, the borrower information section needs to include information such as the company name, unified social credit code, and establishment time; the guarantee information section needs to specify the guarantee type and guarantor identity; and the loan balance information section needs to show data such as loan amount, remaining balance, and overdue status.

[0088] At the same time, the writing specifications corresponding to the sample templates are imported as supplementary context to ensure that the format requirements and expression standards of each section are clearly defined in the large model.

[0089] Extract data and generate reports:

[0090] The target credit data, categorized into business registration data, debt data, and personnel information, is input into the bank's credit business big model. Simultaneously, preset prompts for report generation are imported. These prompts follow a structure of "background + role + input description + task breakdown + output constraints + example." The background is writing a bank credit responsibility assessment report; the role is set as an expert in the field of bank responsibility assessment; the input description explicitly includes JSON structured data and a sample template; the task breakdown involves analyzing the required data items for each section, extracting corresponding data from the JSON (using placeholders if no corresponding data exists), and writing the content for each section according to the template; the output constraints require strict adherence to the sample template format, prohibiting any creative input.

[0091] The large model is based on sample templates and prompt word constraints. First, it extracts the corresponding credit data for each section from the JSON structured data. For example, it extracts the company name and unified social credit code from the business registration data to fill the borrower information section, extracts the guarantee type and guarantor information from the debt data to fill the guarantee information section, and extracts the loan amount, remaining balance and overdue status from the debt data to fill the loan balance information section.

[0092] The content written in each section is then combined and aggregated according to the sample template format to form a complete bank credit responsibility assessment report. The report is written into a Word document and stored with the assistance of Python code, while triggering the report encryption service. For example, a large model dynamically generates random numbers as passwords, Java code completes the document encryption, and pushes the notification to the corresponding employee's APP through the bank's internal interface. The password is only known to the employee.

[0093] In this embodiment, after obtaining the bank's credit liability assessment report, the following may also be included:

[0094] By comparing the consistency between the bank credit liability determination report and the target credit data through an inference-based verification mechanism, eliminating unknown data items and correcting deviations, the final bank credit liability determination report is generated.

[0095] In a specific embodiment, the reasoning verification process is initiated: the reasoning large model is called as the verification core, and the initially generated bank credit responsibility identification report and the corresponding JSON structured target credit data are input into the model to trigger consistency comparison verification.

[0096] Consistency comparison and verification: The inference model is constrained by preset prompt words, which explicitly require that the target credit data be the sole data source. The consistency of all data items in the report is compared with the JSON data, and the report content is checked section by section and field by field.

[0097] Compare data items across different sections, such as borrower information, guarantee information, and loan balance information, including company name, loan amount, and approval personnel information; check whether all data items are from the target credit data and whether there are any newly added unknown data items in the database.

[0098] Verify whether the numerical values ​​and format of the data in the verification report fully match the target credit data, and identify data discrepancies, such as inconsistent units of amount or differences in field descriptions.

[0099] Data correction and optimization:

[0100] If any unknown data items are found in the report that do not have corresponding records in the target credit data, the unknown content will be removed directly.

[0101] If data discrepancies are detected, they should be corrected according to the standards of the target credit data. For example, incorrect monetary units in the report should be standardized to "yuan" and field descriptions that do not match the JSON data should be corrected to ensure that the reported data is completely consistent with the source data.

[0102] Generate the final report: After completing the removal of unknown data items and the correction of deviations, the inference model outputs the final bank credit responsibility determination report that has passed the verification. At the same time, it generates a verification result description, which clearly states that there are no unknown data items and that the data deviations have been corrected, for business personnel to refer to, so as to ensure the accuracy and reliability of the data in the final report.

[0103] The bank credit liability determination report generation method of this invention has been verified to have the following beneficial effects:

[0104] 1. Lower the barrier to entry and make it suitable for non-technical personnel: Based on the prompt word project, natural language and SQL statements are escaped. Business personnel do not need to master Java or other development languages ​​and NLP technology. They only need to enter the customer number and debt number to initiate a request. Non-technical personnel can also operate it easily, which greatly reduces the difficulty of use.

[0105] 2. Enhance versatility and reduce development costs: No customized templates are required. Simply change the data table and report sample templates to reuse them in different credit scenarios. No redesign or development is needed, which significantly reduces the development costs of system iteration and maintenance, and makes the system more adaptable.

[0106] 3. Achieve full-process automation and improve generation efficiency: From data integration, query, and format conversion to report generation and encryption, no manual intervention is required throughout the entire process. The first draft of the responsibility determination report can be generated with one click, solving the problems of long writing cycles and low efficiency caused by traditional reliance on manual sorting.

[0107] 4. Ensure data accuracy and reduce error rate: Through multiple mechanisms such as data cleaning, deduplication, association matching, consistency verification, and large-scale inference model review, combined with the constraint of prompt words on the free play of the large model, we effectively avoid problems such as data disorder, serialization, and unknown data items, significantly reduce the report error rate, and ensure that the data is consistent with the source data.

[0108] This invention also provides a bank credit liability determination report generation device, as described in the following embodiments. Since the principle by which this device solves the problem is similar to the bank credit liability determination report generation method, the implementation of this device can refer to the implementation of the bank credit liability determination report generation method; repeated details will not be elaborated further.

[0109] Figure 4 This is a structural example diagram of the bank credit liability determination report generation device in an embodiment of the present invention, as shown below. Figure 4 As shown, the bank credit liability assessment report generation device includes:

[0110] Data acquisition module 401 is used to acquire bank credit source data, which includes business registration data, debt data, and personnel information;

[0111] The SQL query statement generation module 402 is used to obtain customer ID information and debt number information for credit business, and to obtain the SQL query statement for credit business based on the customer ID information and debt number information for credit business, combined with preset prompt words.

[0112] The target credit data determination module 403 is used to determine the target credit data corresponding to customer number information and debt number information based on the bank credit source data using SQL query statements of credit business.

[0113] The bank credit liability determination report generation module 404 is used to input the target credit data into the bank credit business big model to obtain the bank credit liability determination report; the bank credit business big model is obtained by training a neural network model based on historical bank credit data.

[0114] Figure 5 This is a specific example diagram of the structure of the bank credit liability determination report generation device in an embodiment of the present invention, as shown below. Figure 5 As shown in one embodiment, Figure 4 The bank credit liability determination report generation device shown in the embodiment of the present invention may further include: a data processing module 501.

[0115] In one embodiment, the data processing module 501 is specifically used for:

[0116] After obtaining bank credit source data, the data is cleaned, deduplicated, matched, and standardized in terms of fields.

[0117] In one embodiment, the target credit data determination module 403 is specifically used for:

[0118] Using SQL query statements for credit business, based on customer number information and debt number information, match and extract corresponding business registration data, debt data, and personnel information from bank credit source data;

[0119] The matched business registration data, debt data, and personnel information are checked for data consistency. Once the check is successful, the data is identified as the target credit data.

[0120] Figure 6 This is a specific example diagram of the structure of the bank credit liability determination report generation device in an embodiment of the present invention, as shown below. Figure 6 As shown in one embodiment, Figure 4 The bank credit liability determination report generation device shown in the embodiment of the present invention may further include: a data conversion module 601.

[0121] In one embodiment, the data conversion module 601 is specifically used for:

[0122] After using SQL query statements for credit business to determine the target credit data corresponding to customer number information and debt number information based on bank credit source data, the target credit data is converted into JSON structured data; the JSON structured data is divided into data categories according to business registration data, debt data, and personnel information.

[0123] In one embodiment, the bank credit liability determination report generation module 404 is specifically used for:

[0124] Import a pre-set liability determination report template into the bank's credit business model. The liability determination report template includes a borrower information section, a guarantee information section, and a loan balance information section.

[0125] Input the target credit data into the bank's credit business model to obtain the bank's credit business model based on the preset liability determination report template. Extract the corresponding credit data of each segment from the target credit data to obtain the bank credit liability determination report.

[0126] Figure 7 This is a specific example diagram of the structure of the bank credit liability determination report generation device in an embodiment of the present invention, as shown below. Figure 7 As shown in one embodiment, Figure 4 The bank credit responsibility determination report generation device shown in the embodiment of the present invention may further include: a data verification module 701.

[0127] In one embodiment, the data verification module 701 is specifically used for:

[0128] After obtaining the bank's credit liability assessment report, the consistency between the report and the target credit data is compared through an inferential verification mechanism. Unknown data items are removed and deviations are corrected before the final bank credit liability assessment report is generated.

[0129] Based on the aforementioned inventive concept, such as Figure 8 As shown, the present invention also proposes a computer device 800, including a memory 810, a processor 820, and a computer program 830 stored in the memory 810 and executable on the processor 820. When the processor 820 executes the computer program 830, it implements the aforementioned method for generating a bank credit liability determination report.

[0130] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for generating a bank credit liability determination report.

[0131] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-mentioned method for generating a bank credit liability determination report.

[0132] In this embodiment of the invention, bank credit source data is acquired, including business registration data, debt data, and personnel information; customer ID information and debt ID information for credit business are acquired; based on the customer ID information and debt ID information for credit business, and combined with preset prompt words, an SQL query statement for credit business is obtained; using the SQL query statement for credit business, target credit data corresponding to the customer ID information and debt ID information is determined based on the bank credit source data; the target credit data is input into a large-scale bank credit business model to obtain a bank credit liability determination report; the large-scale bank credit business model is obtained by training a neural network model based on historical bank credit data; this embodiment of the invention, through the fusion of the large-scale model and prompt words, determines a credit report generation scheme that adapts to the characteristics of credit data, balances generation efficiency and accuracy, and lowers the technical threshold for use, based on SQL statements, providing technical support for the automated generation of bank credit liability determination reports.

[0133] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0134] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0135] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

Claims

1. A method for generating a bank credit liability determination report, characterized in that, include: Obtain bank credit source data, which includes business registration data, debt data, and personnel information; Obtain customer ID and debt number information for credit business; based on the customer ID and debt number information for credit business, and combined with preset prompt words, obtain the SQL query statement for credit business. Using SQL query statements for credit business, the target credit data corresponding to customer number information and debt number information is determined based on bank credit source data; The target credit data is input into the bank's credit business model to obtain a bank credit liability determination report; the bank's credit business model is obtained by training a neural network model based on historical bank credit data.

2. The method as described in claim 1, characterized in that, After obtaining bank credit source data, the following is also included: The bank credit source data is cleaned, deduplicated, matched, and standardized in terms of fields.

3. The method as described in claim 1, characterized in that, Using SQL queries for credit operations, target credit data corresponding to customer ID and debt ID information is determined based on bank credit source data, including: Using SQL query statements for credit business, based on customer number information and debt number information, match and extract corresponding business registration data, debt data, and personnel information from bank credit source data; The matched business registration data, debt data, and personnel information are checked for data consistency. Once the check is successful, the data is identified as the target credit data.

4. The method as described in claim 1, characterized in that, After using SQL queries in credit operations to determine the target credit data corresponding to customer ID and debt ID information based on bank credit source data, the process also includes: The target credit data is converted into JSON structured data; the JSON structured data is categorized into business data, debt data, and personnel information.

5. The method as described in claim 1, characterized in that, Inputting the target credit data into the bank's large-scale credit business model yields a bank credit liability determination report, including: Import a pre-set liability determination report template into the bank's credit business model. The liability determination report template includes a borrower information section, a guarantee information section, and a loan balance information section. Input the target credit data into the bank's credit business model to obtain the bank's credit business model based on the preset liability determination report template. Extract the corresponding credit data of each segment from the target credit data to obtain the bank credit liability determination report.

6. The method as described in claim 5, characterized in that, After receiving the bank's credit liability assessment report, the following is also included: By comparing the consistency between the bank credit liability determination report and the target credit data through an inference-based verification mechanism, eliminating unknown data items and correcting deviations, the final bank credit liability determination report is generated.

7. A bank credit liability determination report generation device, characterized in that, include: The data acquisition module is used to acquire bank credit source data, which includes business registration data, debt data, and personnel information. The SQL query statement generation module is used to obtain customer ID information and debt number information for credit business, and to generate SQL query statements for credit business based on the customer ID information and debt number information and preset prompt words. The target credit data determination module is used to determine the target credit data corresponding to customer number information and debt number information based on the bank's credit source data using SQL query statements for credit business. The bank credit liability determination report generation module is used to input target credit data into the bank credit business big model to obtain the bank credit liability determination report; the bank credit business big model is obtained by training a neural network model based on historical bank credit data.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1-6.

Images