Tender review method and apparatus, electronic device, and storage medium

Abstract

This application relates to the field of enterprise management, and more particularly to a method, apparatus, electronic device, and storage medium for bid review. The method involves inputting the bid to be reviewed into a bid review model to generate a review report. The review report includes one or more elements such as serial number, clause location, risk category, risk description, risk level, key verification points, and modification suggestions. Upon the review report meeting preset conditions, the report is sent to a human reviewer for further review of the bid. This application significantly improves review efficiency, accuracy, and standard consistency, while simultaneously ensuring the security of confidential data.

Description

Technical Field

[0001] This application relates to the field of business management, and more particularly to a method, apparatus, electronic device, and storage medium for reviewing tender documents. Background Technology

[0002] Currently, in the process of material procurement, large coal enterprises use technical tender documents as a core component of bidding documents. Their content must strictly comply with national, industry, and internal enterprise standards to ensure the safety, compliance, and feasibility of the procured equipment.

[0003] According to existing policy requirements, large energy companies urgently need to build a safe, controllable, intelligent, and efficient procurement compliance review system. Currently, in the field of technical tender review for corporate procurement, especially for large energy companies (such as coal, electricity, and oil), the mainstream method is still manual, word-by-word review, supplemented by some electronic document comparison tools. Some advanced companies have tried to introduce large-scale model technology, and its typical implementation methods include: 1. Manual review of each clause: The tender documents are reviewed jointly by personnel from multiple departments such as planning, technology, and procurement. The compliance of the clauses, the consistency of parameters, and the standardization of the format are checked word by word. This process is time-consuming and highly subjective. 2. Rule Engine-Assisted Auditing: Some enterprises have deployed rule engine systems based on keyword matching or regular expressions, which can automatically identify " It addresses explicit risks such as "clause number" and "exclusive statements," but cannot handle semantic conflicts, inconsistencies in contextual logic, or cross-document parameter comparisons. 3. General large model auxiliary tools: such as Deepseek, Wenxin Yiyan, and Tongyi Qianwen are used to generate the first draft of the tender document, but they lack the large model to empower the intelligent review of technical tender documents in a specific field, which can easily lead to professional differences and practical application problems on site, and cannot guarantee data security. 4. Third-party cloud platform auditing services: Some companies have tried to upload tender documents to public AI platforms for compliance checks, but due to the involvement of confidential parameters such as the model, power, and size of underground equipment, there is a serious risk of data leakage, which does not meet the requirements of state-owned asset supervision.

[0004] 5. Knowledge Graph-Assisted Review System: Preliminary explorations have been made in industries such as construction and hydropower, where risk warnings are achieved by constructing a knowledge graph of the bidding party, the tendering party, and the project. However, a closed-loop system covering the entire lifecycle of "review-verification-archiving" has not yet been formed. Summary of the Invention

[0005] This application aims to address the challenge of automating, standardizing, and intelligently reviewing enterprise technology bids by utilizing locally deployed large-scale modeling technology and knowledge-enhanced retrieval methods while ensuring the security of classified data. To this end, in its first aspect, this application provides a bid review method, including: Input the tender documents to be reviewed into the tender review model to obtain the review report. The elements of the review report include one or more of the following: serial number, clause location, risk category, risk description, risk level, key points of verification, and modification suggestions. If the audit report meets the preset conditions, the audit report is pushed to a human auditor for review of the tender document.

[0006] Furthermore, the tender document review model is deployed to an intranet environment, which also includes a localized knowledge base. This localized knowledge base is constructed using the following methods: Obtain the basic specification library and historical case library, and generate a custom rule library; A localized knowledge base is built by combining a basic specification library, a historical case library, and a custom rule library.

[0007] Furthermore, after the tender documents to be reviewed are input into the tender document review model, the tender document review model extracts the key information from the tender documents to be reviewed and calls the localized knowledge base to verify the key information.

[0008] Furthermore, key information is verified, including format specification verification, logical consistency verification, compliance verification, and risk classification and tracing.

[0009] Furthermore, after verifying the key information, an audit report is generated based on the verification results.

[0010] Furthermore, the preset condition is that the types and / or quantities of elements contained in the audit report reach a preset threshold.

[0011] Furthermore, after reviewing the tender documents, human auditors will store the corresponding content in the tender documents into the historical case database and update the localized knowledge base.

[0012] In a second aspect, this application provides a tender document review device, comprising: The tender document processing module is used to input tender documents to be reviewed into the tender document review model and generate a review report. The elements of the review report include one or more of the following: serial number, clause location, risk category, risk description, risk level, key points of verification, and modification suggestions. The report push module is used to push the audit report to the human auditor for review when the audit report meets the preset conditions.

[0013] In a third aspect, this application provides an electronic device, including: a processor; and a memory storing a program, wherein the program includes instructions that, when executed by the processor, cause the processor to perform the method described above.

[0014] In a fourth aspect, this application provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the methods described above.

[0015] The beneficial effects of this application are as follows: This application adopts a four-in-one tender document review method, which integrates "localized deployment architecture, knowledge-enhanced retrieval, refined rule engine, and human-machine collaborative hierarchical review". This method achieves full-process automation and intelligence in tender document review, from key information capture and multi-dimensional compliance verification to risk classification report output. It significantly improves review efficiency, accuracy, and standard consistency, while ensuring the security of confidential data. Attached Figure Description

[0016] Figure 1 This is a schematic diagram illustrating the application scenario for the review of the tender documents in this application.

[0017] Figure 2 This is a flowchart illustrating the method for reviewing the bid documents in this application.

[0018] Figure 3 This is a schematic diagram of the bid review device for this application. Detailed Implementation

[0019] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present application, not the entire structure.

[0020] This application provides a specific implementation example of a tender document review method, such as... Figure 1 As shown, this application scenario includes a tender document 100 to be reviewed, an enterprise intranet 101, a tender document review model 102, a localized knowledge base 103, and a human reviewer 104. The tender document review model 102 and the localized knowledge base 103 are deployed on the enterprise intranet 101. When the tender document review model 102 receives a tender document 101 to be reviewed, it calls the localized knowledge base 103 to review the tender document 100 and generates a review report. When the review report meets preset conditions, it is pushed to the human reviewer 104.

[0021] The tender document review method provided in this application embodiment can be executed by the tender document review device provided in this application embodiment, or by a computer device that integrates the tender document review method.

[0022] The first aspect of this application provides a method for reviewing tender documents, the process of which is as follows: Figure 2 As shown, it includes: S201 inputs the tender documents to be reviewed into the tender review model to obtain an review report. The elements of the review report include one or more of the following: serial number, clause location, risk category, risk description, risk level, key points of verification, and modification suggestions. S202 responds to the audit report meeting the preset conditions and pushes the audit report to the human auditor for review of the tender document.

[0023] In step S201, the tender review model can be a pre-trained large language model. After the tender to be reviewed is input into the tender review model, the tender review model processes the tender to be reviewed and obtains the review report.

[0024] To ensure the bid review model is adapted to the professional field of the bids to be processed, the model calls upon a localized knowledge base to process the bids. Considering that bids are classified documents with high data security risks, this application deploys the bid review model and localized knowledge base within the enterprise intranet environment to meet the confidentiality requirements of bid processing.

[0025] To enhance the processing capabilities of the tender document review model, this application proposes a method for constructing a localized database, including: Obtain the basic specification library and historical case library, and generate a custom rule library; By adopting an incremental update and full verification mechanism, a localized knowledge base is built by combining the basic specification library, historical case library, and custom rule library, and a blockchain evidence storage operation log is generated.

[0026] Basic Standards Library: Includes national standards (GB), industry standards (MT, AQ, etc.), enterprise internal standards (such as "Technical Tender Compilation and Review Standards and Specifications"), coal mine safety regulations, various material manuals and papers, etc., and is stored in a structured format as "Clause ID-Content-Applicable Scenario-Risk Level".

[0027] Historical Case Library: Includes 1000+ historical audit cases, marked with error types (such as "parameter contradiction", "outdated reference", "exclusive clause").

[0028] Custom rule base: Allows administrators to upload PDF / TXT / DOCX files or web links, which are automatically extracted and structured into audit rules using NLP technology.

[0029] The method for constructing a localized database as described in this application ensures the timeliness and flexibility of knowledge, dynamically optimizes the localized database, and meets the traceability requirements of state-owned asset supervision.

[0030] In step S201, the tender document review model processes the tender documents to be reviewed, including key information extraction, multi-dimensional verification, and generation of a structured review report.

[0031] When extracting key information, OCR+NLP technology is used to parse the document structure of the tender document to be reviewed, and to extract reference models, core parameters, etc. Clause No. 1, Acceptance Standards.

[0032] Multi-dimensional verification includes format specification verification, logical consistency verification, compliance verification, and risk classification and tracing.

[0033] Format specification verification includes: Inspection mark Does the clause account for ≤30%? Check if the parameters are set to a range value (e.g., "Power: ≥18kW" instead of "Power: 20kW"). Check for formatting errors such as units of measurement, punctuation, and spaces.

[0034] Logical consistency checks include: Compare whether the "Goods Requirements List" matches the "Technical Parameters" (e.g., "Inconsistency between equipment weight and size"). Check for any contradictory statements between the "procurement requirements" and the "technical specifications" (e.g., "explosion-proof required" but "explosion-proof standards not cited").

[0035] Compliance verification includes: By using RAG technology, keyword search (BM25) and vector search (such as Sentence-BERT) are combined to match the latest standards; Identify risks such as "referencing outdated specifications", "exclusivity clauses" (such as "designating XX brand"), and "whether it will lead to patent disputes".

[0036] Risk classification and source tracing include: The problem is classified into five risk levels: severe, high, medium, low, and minor.

[0037] Through the above multi-dimensional verification, the processing capability of the tender document processing model has been systematically improved, and the rigor and compliance of the model verification capability have been comprehensively enhanced.

[0038] When generating a structured audit report, the tender audit model outputs potential problems in the tender document according to a preset format. This implementation example uses a preset format, which includes: serial number, clause location, risk category, risk description, risk level, key points for verification, and modification suggestions.

[0039] In step S202, considering that the content involved in the review report may require complex decisions, when the review report meets the preset conditions, the review report is pushed to a human reviewer to review the content that requires complex decisions, so as to avoid the tender review model from misunderstanding the implicit specified clauses or special equipment compatibility.

[0040] For example, when a serious or high-risk issue is found in the audit report, the relevant content in the audit report is sent to a human auditor for review, while other content is directly archived.

[0041] This application takes into account that the content requiring complex decisions is usually new knowledge brought about by technological iteration. Therefore, the results of human review will also be included in the historical case library, and the localized knowledge base will optimize the verification rules through incremental learning.

[0042] This application employs a four-pronged approach to bid document review: localized deployment architecture, knowledge-enhanced retrieval, refined rule engine, and human-machine collaborative layered review. This approach automates and intelligently manages the entire process of bid document review, from key information extraction and multi-dimensional compliance verification to risk classification report output, significantly improving review efficiency, accuracy, and standard consistency while ensuring the security of confidential data. Based on a general-purpose large language model and combined with a localized knowledge base for knowledge-enhanced retrieval, a dedicated large-model intelligent review system for technical bid documents has been constructed.

[0043] This application further discloses some technical effect data, compared with the prior art: 1. Efficiency increased by 70%: The review time for a single document was reduced from 4-6 hours to 30-60 minutes.

[0044] 2. Significantly improved quality: The rate of intercepting hidden risks exceeded 95%, and exclusive clauses such as "designated brand" were successfully identified; the compliance rate increased from 80% to 95%; and low-level errors such as model conflicts and parameter inconsistencies were eliminated.

[0045] 3. Improved standardization: The consistency of review conclusions for similar documents has increased from 60% to 95%, reducing redundant communication between departments.

[0046] 4. Data security and controllability: The entire system is deployed on the enterprise intranet, and confidential parameters are not transmitted to external parties, meeting the requirements of state-owned asset supervision.

[0047] 5. High efficiency through human-machine collaboration: Machines handle 70% of standardized verification tasks, while humans focus on 30% of complex decisions, balancing efficiency and quality.

[0048] A second aspect of this application provides a tender document review device. Figure 3 As shown, the tender document review device can be integrated into an electronic device, specifically including a tender document processing module and a report push module.

[0049] The tender document processing module is used to input tender documents to be reviewed into the tender document review model and generate a review report. The elements of the review report include one or more of the following: serial number, clause location, risk category, risk description, risk level, key points of verification, and modification suggestions. The report push module is used to push the audit report to the human auditor for review when the audit report meets the preset conditions.

[0050] The tender document review device provided in this application includes a tender document processing module and a report push module whose functions correspond one-to-one with the steps in the tender document review method described above. For a detailed explanation of the tender document review device and related refinements and optimizations, please refer to the specific embodiments in the tender document review method described above, which will not be repeated here.

[0051] In some embodiments, this application also provides an electronic device, which may be a mobile phone, computer, or tablet computer, including a memory and a processor. The memory stores a calculator program, which, when executed by the processor, implements the tender review method in the above embodiments.

[0052] The processor is used to execute all or part of the steps of the tender review method in the above embodiments. The memory is used to store various types of data, which may include, for example, instructions for any application or method in the electronic device, as well as application-related data.

[0053] The processor can be implemented as an Application Specific Integrated Circuit (ASIC), Digital Signal Processor (DSP), Digital Signal Processing Device (DSPD), Programmable Logic Device (PLD), Field Programmable Gate Array (FPGA), controller, microcontroller, microprocessor, or other electronic components, and is used to execute the steps of the tender review method in the above embodiments.

[0054] Memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0055] In some embodiments, this application also provides a computer-readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, server, App application store, etc., which stores a computer program. When the computer program is executed by a processor, it can implement the above method steps. For specific implementation processes, please refer to the above embodiments. This embodiment will not be repeated here.

[0056] It should be understood that although the steps in the flowcharts of the various embodiments of this application are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0057] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

[0058] The above are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method of bid review, the method comprising: include: Input the tender documents to be reviewed into the tender review model to obtain the review report. The elements of the review report include one or more of the following: serial number, clause location, risk category, risk description, risk level, key points of verification, and modification suggestions. In response to the audit report meeting preset conditions, the audit report is pushed to a human auditor for review of the tender document.

2. The method of claim 1, wherein, The tender document review model is deployed to an intranet environment, which also includes a localized knowledge base. The localized knowledge base is constructed using the following methods: Obtain the basic specification library and historical case library, and generate a custom rule library; A localized knowledge base is constructed using the aforementioned basic specification library, the aforementioned historical case library, and the aforementioned custom rule library.

3. The method of claim 2, wherein, After the tender documents to be reviewed are input into the tender document review model, the tender document review model extracts the key information from the tender documents to be reviewed and calls the localized knowledge base to verify the key information.

4. The method of claim 3, wherein, The key information is verified, including format specification verification, logical consistency verification, compliance verification, and risk classification and tracing.

5. The method of claim 4, wherein, After verifying the key information, the audit report is generated based on the verification results.

6. The method of claim 1, wherein, The preset condition is that the types and / or quantities of elements contained in the audit report reach a preset threshold.

7. The method of claim 2, wherein, After reviewing the tender documents, human reviewers store the corresponding content in the tender documents into the historical case database and update the localized knowledge base.

8. A bid review apparatus, characterized by, include: The tender document processing module is used to input the tender documents to be reviewed into the tender document review model and obtain the review report. The elements of the review report include one or more of the following: serial number, clause location, risk category, risk description, risk level, key points of verification, and modification suggestions. The report push module is used to push the audit report to a human auditor for review of the tender documents in response to the audit report meeting preset conditions.

9. An electronic device, comprising: include: processor; And a memory storing a program, wherein the program includes instructions that, when executed by the processor, cause the processor to perform the method as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method as described in any one of claims 1-7.

Images