Automatic bid invitation file auditing system based on Java rule engine

Abstract

The invention discloses an intelligent procurement auditing system and method based on a multi-modal technology, and the system collects the text and image data of a procurement page in real time through a browser plug-in, fuses text semantic features and image visual features into multi-modal comprehensive features through the multi-modal preprocessing and feature fusion technology, and carries out the auditing of the procurement page. A pre-training large model is implanted in a browser plug-in mode, and compliance auditing is carried out in combination with purchasing rules. The system does not need to be in interface butt joint with a purchasing system, and real-time checking and early warning of the risk in the purchasing process can be achieved. In practical application, the complex rule recognition accuracy can reach 95% or above, the auditing efficiency and accuracy are remarkably improved, and the method is suitable for intelligent auditing and risk management and control of the enterprise purchase process.

Description

Technical Field

[0001] This invention relates to the field of computer application technology, and in particular to an automated tender document review system based on a Java rule engine. Background Technology

[0002] In current tender document review processes, the traditional manual review model has many drawbacks that urgently need to be addressed. The review process is often extremely cumbersome, requiring reviewers to compare each clause in the tender document item by item, making it difficult to meet the rapid processing needs of large-scale tender documents. Manual review is affected by fatigue and differences in experience, resulting in a missed inspection rate of over 15%, especially in complex clauses and lengthy documents, where subjective judgment can easily lead to oversights. Different reviewers may have different understandings of the same clause, leading to inconsistent review results.

[0003] Existing electronic systems are mostly based on simple rule-matching algorithms, lacking the ability to understand complex logic and natural language, and unable to handle ambiguous expressions or cross-clause relevance issues. Traditional systems struggle to identify semantic ambiguities or potential risks in clauses, leading to inaccurate review results. Existing systems cannot dynamically adapt to rule changes and lack the ability to respond quickly to new industry standards or policy adjustments.

[0004] Different industries have vastly different requirements for technical parameters and business terms, and traditional systems lack the ability to dynamically configure rules. Existing systems are mostly custom-developed for specific industries, making it difficult to quickly adapt to the rule requirements of other industries, resulting in poor system scalability. Rule updates rely on code modifications, which are inflexible and unable to quickly respond to changes in industry standards or policies, increasing project compliance risks.

[0005] The tender documents contain a large amount of unstructured data (such as scanned PDFs, nested tables, handwritten annotations, etc.). Traditional OCR tools have an accuracy rate of less than 70% in recognizing complex formats, making data extraction and verification difficult. Existing OCR tools cannot directly verify the compliance of unstructured data, requiring secondary manual verification, which increases the auditing burden. Traditional systems have limited capabilities in processing complex content such as nested tables and handwritten signatures, making it difficult to guarantee data integrity and accuracy.

[0006] When multiple reviewers in different locations modify documents concurrently, conflicting clauses often arise, leading to inconsistent document versions. Traditional version control tools cannot detect these conflicts in real time and lack granular recording and auditing capabilities for modification history, making it difficult to trace the source of the problem. This chaotic version control results in inconsistent review outcomes, increases project coordination costs, and may even trigger legal disputes.

[0007] Some systems are developed using C# or Python, which presents cross-platform compatibility issues, especially in multi-operating system environments (such as Windows, Linux, and macOS), resulting in high deployment and maintenance costs. Systems implemented using non-Java technologies have security vulnerabilities; for example, C# systems running on Linux rely on additional compatibility layers, increasing the risk of attacks. Traditional system architectures are closed, requiring specialized developers for rule updates and feature expansions, leading to high maintenance costs and poor scalability.

[0008] Based on the above situation, there is an urgent need for an automated bidding document review system with high processing capabilities, intelligent review mechanisms, and flexible adaptability to completely solve the inherent defects of the traditional review model. Summary of the Invention

[0009] 1. Rule Management Module: Features a built-in multi-industry, multi-scenario audit rule library, supporting dynamic configuration and updates. Rules are stored based on the DM database and can be managed by industry, project category, and procurement method. The Drools rule engine enables dynamic loading and execution of rules, automatically triggering validation during document preparation and review, synchronously marking non-compliant items and providing modification suggestions.

[0010] 2. Contract Management Module: Provides a multi-industry, multi-scenario contract template library, supporting template filtering, editing, and reuse. Templates are stored using MongoDB and feature version management and metadata annotation. A front-end filter enables quick searching by industry, project category, and other dimensions, and supports strong binding with tender document templates to ensure consistency of terms.

[0011] 3. Tender Document Template Management Module: Covers a template library encompassing multiple industries and project categories, supporting creation, editing, and version management. Template content can be edited online with embedded rule validation logic; updates automatically trigger validation to ensure template matching with rules; historical version tracking and difference comparison are achieved based on a version control system.

[0012] 4. Intelligent Document Compilation Module: Supports automated compilation of tender documents, integrates Office file processing capabilities, and enables version tracking and format verification. It generates Word, Excel, and other format files based on the Apache POI library; it automatically saves versions after modifications, supports historical version comparison and recovery; and verifies file formats using regular expressions and XML Schema to ensure compliance with industry standards.

[0013] 5. Intelligent Review Module: After a file is modified, the review process is automatically or manually triggered. The system performs secondary verification based on a rule engine to avoid missing errors and generates a detailed review report containing problem items and modification suggestions. Attached Figure Description

[0014] Figure 1 This is a diagram of the overall architecture of the present invention. Figure 2 Flowchart for preparing tender documents for this invention Figure 3 This is a flowchart illustrating the contract template binding process of the present invention. Figure 4 This is a flowchart of the intelligent review process of the present invention. Detailed Implementation

[0015] 1. System Setup: The system is developed using the Java language platform and employs the Spring Boot architecture to build the basic framework, achieving decoupling and efficient interaction between the presentation layer, business logic layer, data access layer, and data storage layer. The presentation layer uses Vue.js to develop a responsive web interface, supporting adaptive form rendering and multi-terminal data synchronization; the data access layer is adapted to the DM database, uses MyBatis for interaction with a domestic database, and integrates Redis for caching frequently accessed data; Spring Security is used to implement RBAC-based access control, ensuring clear boundaries for different roles and meeting system security requirements.

[0016] 2. Rule Engine Implementation: Built on the Drools rule engine, it adopts a three-layer mapping structure of "clause-rule-parameter": the bottom layer connects to the regulatory database, the middle layer uses a visual configuration platform (supporting drag-and-drop logic orchestration) to convert regulatory clauses into executable rules, and the top layer sets dynamic parameters (such as qualification validity thresholds and material supplementation deadlines). The rule engine supports version management and conflict detection. When a new industry standard is released, the system automatically marks the rules that need to be updated and pushes revision suggestions. After the reviewers complete the rule adjustments through the interface, the changes take effect immediately without restarting the system.

[0017] 3. Intelligent Documentation Module Implementation: Integrates the Apache POI library to automate the processing of Office documents, supporting the generation and editing of Word, Excel, and other file formats. Each modification automatically triggers a version saving mechanism, enabling historical version tracking and difference comparison through a version control system. File format validation is implemented based on regular expressions and XML Schema to ensure that the tender documents comply with industry standards and format requirements.

[0018] 4. Blockchain Evidence Storage Implementation: Deploy consortium blockchain nodes to store key modification records of bidding documents on the blockchain, including material hash values, approval conclusions, timestamps, digital signatures, and other information. Leveraging the immutability of blockchain, the transparency and traceability of the review process are ensured, meeting audit requirements.

[0019] 5. System Workflow: Tender Document Preparation Process: Users select the industry, project category, and procurement method through the system, and the system automatically matches the corresponding templates and rules; users edit the document content online, and the system verifies the format and clause compliance in real time, marks problematic items, and provides modification suggestions; after the document is prepared, the system automatically generates a version record and triggers the review process.

[0020] Contract template binding process: After the user selects a tender document template, the system automatically recommends a matching contract template; the user can edit the contract content online, and the system automatically verifies the consistency between the contract terms and the tender requirements.

[0021] Audit traceability process: The system records the modification history, review path, and rule reference basis of each clause, and generates blockchain evidence files; auditors can query historical versions and evidence files through the system to ensure the transparency of the document preparation and review process.

Claims

1. A Java rule engine based automated bid document review system, characterized in that, Comprises: Rule management module: built-in multi-industry and multi-scenario audit rule library (DM database storage), dynamically loaded and executed by Drools engine, automatically checked and marked for non-compliance items during preparation and audit stages, and provided with modification suggestions; Contract management module: contains multi-industry and multi-scenario contract template library (MongoDB storage), supports version management and metadata annotation, and can be quickly found and strongly bound with the bidding document template through front-end filtering; Bidding document template management module: contains multi-industry and multi-project category template library, supports creation, editing and version management, can be edited online and embedded with verification logic, and automatically triggers verification after update, realizes historical tracking and difference comparison through version control system; Intelligent preparation module: supports automatic preparation, built-in Office processing function (based on ApachePOI to generate Word / Excel), automatically stores version after modification, supports version comparison and recovery, and checks format through regular expressions and XMLSchema; Intelligent review module: automatically triggers review after file modification, secondarily checks through rule engine, and generates detailed report containing problem items; Blockchain storage module: records clause modification history, audit path and rule basis, and generates tamper-proof storage file.

2. The system of claim 1, wherein: The rule management module supports visual configuration and dynamic update of rules (period reduced to several hours), which can be classified by industry, project category and procurement method.

3. The system of claim 1, wherein: The intelligent preparation module integrates OCR technology, supports identification and extraction of unstructured data such as scanned PDF and nested tables, and the accuracy rate is over 70%.

4. The system of claim 1, wherein: The blockchain storage module deploys alliance chain nodes, and the storage contains material hash value, approval conclusion, etc., and supports full-process tracking.

5. The system of claim 1, wherein: Based on Java development, using Spring Boot architecture, supporting multiple systems, integrating Spring Security to realize RBAC permission control.

6. An auditing method based on any of the systems of claims 1-5, characterized in that, Comprises: Receives user-selected industry information, etc., automatically matches templates and rules; Receives online editing content, checks compliance in real time and marks problems, and provides suggestions; After preparation, automatically generates version records and triggers review; after secondary verification by rule engine, generates detailed review report; records clause modification history, etc. to blockchain storage file.

7. The method of claim 6, wherein: Real-time verification uses Drools engine, supports dynamic loading and unstructured data OCR identification.

8. The method of claim 6, wherein: Blockchain storage contains material hash value, approval conclusion, etc., ensuring tamper resistance.

Images