Failure point one-key traceability method in intelligent text auditing system

Abstract

The invention discloses a failure point one-key tracing method in an intelligent text auditing system, and the method specifically comprises the following steps: S1, constructing an intelligent text auditing system which comprises an auditing rule configuration module and a rule engine, and the auditing rule configuration module carries out the configuration of a corresponding auditing rule according to the type of a to-be-audited text; the auditing engine calls a corresponding auditing rule according to the to-be-audited text type; s2, inputting a to-be-audited text, analyzing the audited text, inputting the analyzed to-be-audited text into a rule engine, and processing and comparing the input text by the rule engine; and S3, the rule engine outputs a processing comparison result, and auditing of the text data is completed. According to the method and the device, the specific position and the reason of the rechecking failure rule point of the rechecking task can be quickly positioned, the automation rate of the rechecking work is improved, and the efficiency and the quality of manual intervention of business personnel are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a method for one-key tracing failure points in an intelligent text auditing system. Background Art

[0002] The intelligent text review system is an audit business system built based on cutting-edge artificial intelligence technology and professional audit knowledge in various fields, which assists various business personnel to complete complex text review work; the system adopts large language model technology, original rule engine and audit engine technology to assist in completing a series of tasks such as review, archiving, and analysis. It helps customers control risks, improve efficiency, reduce costs, and optimize the audit process.

[0003] At present, traditional text audit systems can generally only provide the reasons for the failure of the audit points, but cannot quickly locate the specific errors in the original text data and original attachment data. When the original text data is large and the attachment data is very complex, it will take a lot of time to find the specific error points and manually intervene, which will seriously affect the audit efficiency and audit quality. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the above-mentioned prior art and to provide a method for one-click tracing of failure points in an intelligent text audit system. The method of the present invention can locate the failure points of original text data and original attachment data by one-click tracing based on extraction and positioning of extraction models, thereby realizing the rapid positioning of the specific location and cause of the failure rule points of the audit task re-audit, thereby improving the automation rate of the audit work and improving the efficiency and quality of manual intervention by business personnel.

[0005] In order to achieve the above invention purpose, the technical solution provided by the present invention is as follows:

[0006] A method for one-click tracing failure points in an intelligent text review system, the method specifically comprising the following steps:

[0007] S1, build an intelligent text review system. The review system includes a review rule configuration module and a rule engine. The review rule configuration module configures the corresponding review rules according to the type of text to be reviewed; the review engine calls the corresponding review rules according to the type of text to be reviewed;

[0008] S2, input the text to be reviewed, parse the review text, input the parsed text to be reviewed into the rule engine, and the rule engine processes and compares the input text;

[0009] S3, the rule engine outputs the processing comparison results and completes the review of the text data.

[0010] Furthermore, the parsing process of the text to be reviewed is specifically: parsing the text to be reviewed into text data and attachment data, wherein the text data is the document data in the text to be reviewed, and the attachment data includes the bills, documents and contracts in the text to be reviewed.

[0011] Furthermore, the text data is directly input into the rule engine, and the attachment data is first input into the AI ​​engine for processing and then into the rule engine. The AI ​​engine processes the attachment data specifically by: parsing the document title, paragraph parsing, field extraction, field position analysis, character recognition, character color, character position and background color in the attachment data.

[0012] Furthermore, the rule engine processes and compares the input text specifically as follows: the rule engine compares the text data with the parsed attachment data one by one. When the information in the text data is inconsistent with the information in the attachment data, the corresponding position in the attachment data is highlighted, and the corresponding text data and attachment data are displayed and compared.

[0013] Furthermore, the text data and attachment data of the text to be reviewed after being de-processed are stored in a system database for use in subsequent review by a rule engine.

[0014] Based on the above technical solution, the one-key tracing failure point method in the intelligent text review system of the present invention has achieved the following technical advantages through practical application:

[0015] 1. The present invention provides a method for one-click tracing failure points in an intelligent text audit system, which can locate the failure points of original text data and original attachment data through one-click tracing based on extraction and positioning by extraction model, and can realize the rapid positioning of the specific position and cause of the failure rule point of the audit task re-audit, thereby improving the automation rate of the audit work and improving the efficiency and quality of manual intervention by business personnel. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is described below by specific examples. However, it should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0017] Sometimes there are dozens or even hundreds of invoices in an audit task. When checking invoices, the traditional text audit method only prompts that the invoice verification has failed, but cannot jump to the corresponding failed invoice with one click on the interface, nor can it quickly locate which field of which invoice failed to be checked. The one-click tracing and locating failure point method based on extraction and positioning proposed in the present invention can not only prompt the specific reason for the audit failure, but also quickly locate the specific erroneous bill with one click, and highlight the erroneous field for manual quick verification.

[0018] During the audit task, multiple contracts with hundreds or even thousands of pages will also be audited. Taking the field check of Party A in the contract as an example, the traditional text audit method only prompts that the consistency check of Party A in the contract has failed in the audit of multiple contracts with hundreds of pages, but cannot give a specific location. In this way, manual search is required in multiple long document contracts one by one, which leads to low audit efficiency. The method proposed in the present invention can complete the contract and field location in multiple long contracts, greatly improving the audit efficiency.

[0019] The present invention provides a method for one-key tracing failure points in an intelligent text audit system, the method specifically comprising the following steps:

[0020] S1, build an intelligent text review system. The review system includes a review rule configuration module and a rule engine. The review rule configuration module configures the corresponding review rules according to the type of text to be reviewed; the review engine calls the corresponding review rules according to the type of text to be reviewed;

[0021] S2, input the text to be reviewed, parse the review text, input the parsed text to be reviewed into the rule engine, and the rule engine processes and compares the input text;

[0022] S3, the rule engine outputs the processing comparison results and completes the review of the text data.

[0023] The parsing process of the text to be reviewed specifically includes: parsing the text to be reviewed into text data and attachment data, wherein the text data is the document data in the text to be reviewed, and the attachment data includes the bills, documents and contracts in the text to be reviewed.

[0024] The text data is directly input into the rule engine, and the attachment data is first input into the AI ​​engine for processing and then into the rule engine. The AI ​​engine processes the attachment data specifically by: parsing the document title, paragraph parsing, field extraction, field position analysis, character recognition, character color, character position and background color in the attachment data.

[0025] The rule engine processes and compares the input text specifically as follows: the rule engine compares the text data with the parsed attachment data one by one. When the information in the text data is inconsistent with the information in the attachment data, the corresponding position in the attachment data is highlighted, and the corresponding text data and attachment data are displayed and compared.

[0026] The text data and attachment data after the text to be reviewed are stored in the system database for subsequent review by the rule engine.

[0027] The rule engine in the intelligent text audit system of the present invention will receive two original data: the original text data and the original attachment data after being parsed by the AI ​​engine. The rule engine will complete the audit of the original data one by one according to the configured audit rules; when an error occurs at a certain audit point, the audit engine will analyze and give the reason for the failure. The audit engine will display the original attachment structured data (voucher) and the original text data (original order) involved, among which the failure reason and the original attachment structured data both support clicks, and after clicking, you can jump to the failed attachment with one click; at the same time, the attachments involved will also be highlighted and flashing, which is convenient for manual rapid positioning and intervention.

[0028] Example

[0029] Taking the review of business trips in travel reimbursement as an example, the financial audit manager sets the review points such as the date, amount, and personnel for reviewing the invoice in the audit rule configuration module according to the financial audit requirements. When the front desk billing personnel submit the bill, they will edit the travel details, such as the date of travel, the passenger, the amount, etc., into text data and input them into the bill system, and then upload the invoice attachments (PDF, pictures) into the bill system. Subsequently, the text data in the bill information will be directly connected to the rule engine, and the AI ​​engine will parse the attachment data uploaded by the bill person and transmit it to the rule engine. The rule engine will compare the above text data with the attachment data and display the comparison results. In the above large process, there are often a lot of bill data and a lot of invoice attachments uploaded. Sometimes, there may be dozens or even hundreds of invoices in actual business scenarios. In this case, if the audit fails, the rule engine can prompt the financial auditor with the specific failure data in the corresponding bill data and the field of which invoice has a problem in the invoice image, which greatly accelerates the work efficiency of the financial auditor.

[0030] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, ordinary technicians in the field should understand that the specific implementation methods of the invention can still be modified or some technical features can be replaced by equivalents without departing from the spirit of the technical solution of the present invention, which should be included in the scope of the technical solution for protection of the present invention.

Claims

1. A method for one-click tracing of failure points in an intelligent text review system. It is characterized in that The method specifically comprises the following steps: S1, build an intelligent text review system. The review system includes a review rule configuration module and a rule engine. The review rule configuration module configures the corresponding review rules according to the type of text to be reviewed; the review engine calls the corresponding review rules according to the type of text to be reviewed; S2, input the text to be reviewed, parse the review text, input the parsed text to be reviewed into the rule engine, and the rule engine processes and compares the input text; S3, the rule engine outputs the processing comparison results and completes the review of the text data.

2. According to claim 1, a method for tracing failure points with one key in an intelligent text audit system, It is characterized in that The parsing process of the text to be reviewed specifically includes: parsing the text to be reviewed into text data and attachment data, wherein the text data is the document data in the text to be reviewed, and the attachment data includes the bills, documents and contracts in the text to be reviewed.

3. According to claim 2, a method for tracing failure points with one key in an intelligent text audit system, It is characterized in that The text data is directly input into the rule engine, and the attachment data is first input into the AI ​​engine for processing and then into the rule engine. The AI ​​engine processes the attachment data specifically by: parsing the document title, paragraph parsing, field extraction, field position analysis, character recognition, character color, character position and background color in the attachment data.

4. According to claim 3, a method for tracing failure points with one key in an intelligent text audit system, It is characterized in that The rule engine processes and compares the input text specifically as follows: the rule engine compares the text data with the parsed attachment data one by one. When the information in the text data is inconsistent with the information in the attachment data, the corresponding position in the attachment data is highlighted, and the corresponding text data and attachment data are displayed and compared.

5. According to claim 1, a method for tracing failure points with one key in an intelligent text audit system, It is characterized in that The text data and attachment data after the text to be reviewed are stored in the system database for subsequent review by the rule engine.