An AI large model-based bidding data evaluation method

Abstract

The application provides an AI large model-based bidding data evaluation method, which comprises the following steps: obtaining a bid document of a target bidding project, and extracting a plurality of historical bidding documents associated with the target bidding project from a historical database of an electronic bidding platform; analyzing the bid document by an AI large model to extract evaluation indexes of at least one evaluation dimension; generating a matching degree score of each historical bidding document under the evaluation indexes by the AI large model; selecting at least one reference bidding document from the plurality of historical bidding documents according to the matching degree score, determining at least one virtual bidding scheme based on the reference bidding document and the evaluation indexes; and generating an evaluation report for the target bidding project by the AI large model according to the matching degree score and the virtual bidding scheme. The application can provide multi-dimensional strategy prediction and optimization suggestions for the bidding party, and significantly improves the intelligent level and decision reference value of bidding evaluation.

Description

Technical Field

[0001] This application relates to the fields of data processing and artificial intelligence technology, and more specifically, to a method for evaluating bidding data based on a large AI model. Background Technology

[0002] Bidding and tendering activities are an important transaction method in a market economy. In the traditional bidding and tendering process, the tendering party needs to review the bids from numerous bidders according to the tender requirements in order to select the best winning bidder. This process usually relies on the subjective experience and professional knowledge of the review experts, which has problems such as low review efficiency, strong subjectivity, and difficulty in comprehensively comparing historical data.

[0003] With the development of artificial intelligence technology, some existing technologies attempt to use machine learning models to assist in the analysis of bid documents. However, existing solutions can usually only score a single bid document or simply analyze historical winning bid data. They cannot combine the specific requirements of the bidding project to deeply match the historical bidding behavior of potential bidders, nor can they proactively generate valuable virtual bidding schemes for the bidding party to make strategic predictions. Therefore, there is an urgent need for a bidding data evaluation method that can fully utilize historical data and combine it with the generation capabilities of large AI models. Summary of the Invention

[0004] The present invention aims to overcome the shortcomings of the prior art and provide a bidding data evaluation method based on AI large model to solve the technical problems of low bidding evaluation efficiency and lack of prediction and generation capabilities in the prior art.

[0005] In a first aspect, the present invention provides a method for evaluating bidding data based on an AI large-scale model, applied to an electronic bidding platform, the method comprising: In response to an evaluation instruction for a target tender project, the tender documents for the target tender project are obtained, and multiple historical tender documents associated with the target tender project are extracted from the historical database of the electronic tendering platform. The tender document and the multiple historical tender documents are input into a pre-built AI big model. The AI ​​big model parses the tender document and extracts evaluation indicators for at least one evaluation dimension. Using the AI ​​big data model and based on the evaluation indicators, the matching degree of each historical bid document is calculated, and a matching degree score of each historical bid document under the evaluation indicators is generated. Based on the matching score, at least one reference bid document is selected from the multiple historical bid documents, and at least one virtual bid scheme is determined by the AI ​​big data model based on the reference bid document and the evaluation index. Based on the matching score and the virtual bidding scheme, the AI ​​big data model is used to generate an evaluation report for the target bidding project.

[0006] Preferably, the evaluation dimensions include at least the technical solution dimension, the business quotation dimension, and the enterprise qualification dimension; The evaluation indicators include key technical points corresponding to the technical solution dimension, price range corresponding to the business quotation dimension, and qualification level requirements corresponding to the enterprise qualification dimension.

[0007] Preferably, using the AI ​​big data model, based on the evaluation indicators, a matching degree calculation is performed on each of the historical bid documents to generate a matching degree score for each of the historical bid documents under the evaluation indicators, including: Using the AI ​​big model, the technical matching degree of each of the historical bid documents is calculated under the technical solution dimension, the price matching degree under the business pricing dimension, and the qualification matching degree under the enterprise qualification dimension. Based on a preset weight allocation, the matching degree of technology, the matching degree of price, and the matching degree of qualifications are weighted and summed to determine the matching degree score.

[0008] Preferably, obtaining the tender documents for the target tender project and extracting multiple historical tender documents associated with the target tender project from the historical database of the electronic bidding platform includes: Obtain the project type and project scale of the target bidding project; Based on the project type and the project size, extract historical bid documents from the historical database that have the same project type and a project size within a preset similarity range.

[0009] Preferably, the step of parsing the tender document using the AI ​​big data model to extract evaluation indicators for at least one evaluation dimension includes: The natural language processing module in the AI ​​model is used to perform text segmentation and semantic recognition on the tender document to identify the key requirement fields in the tender document; The key requirement fields are mapped to a preset evaluation indicator system to generate the structured evaluation indicators.

[0010] Preferably, the step of generating at least one virtual bidding scheme based on the reference bid documents and the evaluation indicators using the AI ​​big data model includes: The reference tender document is used as a sample to be input into the AI ​​large model; Using the generative module of the AI ​​big model and the evaluation indicators, parameter mutations and strategy optimizations are performed on the reference bid documents to generate one or more virtual bid proposals that differ from the reference bid documents in terms of technical solutions and commercial quotations.

[0011] Preferably, the step of generating an evaluation report for the target bidding project using the AI ​​big data model based on the matching score and the virtual bidding scheme includes: The selected reference bid documents and their corresponding matching scores, along with the generated virtual bidding scheme, are used as comparative analysis data and input into the AI ​​big model. The generative module of the AI ​​big data model, combined with the evaluation indicators, generates an evaluation report that includes predictions of the probability of winning the bid, suggestions on pricing strategies, suggestions for optimizing technical solutions, and warnings of potential risks.

[0012] Preferably, it further includes: In response to user feedback on the electronic bidding platform, obtain user feedback data on the evaluation report; The feedback data is used as fine-tuning training data to optimize the parameters of the large AI model, thereby updating the large AI model.

[0013] Preferably, the AI ​​big model is a pre-trained big language model, and the big language model has been fine-tuned in the domain using historical bidding text corpus in the electronic bidding platform.

[0014] In a second aspect, the present invention provides a readable medium including executable instructions, which, when executed by a processor of an electronic device, cause the electronic device to perform any of the methods described in the first aspect.

[0015] Thirdly, the present invention provides an electronic device including a processor and a memory storing execution instructions, wherein when the processor executes the execution instructions stored in the memory, the processor performs any of the methods described in the first aspect.

[0016] This invention provides a bidding data evaluation method based on an AI-powered big data model. The method uses the AI ​​model to deeply analyze the tender documents of a target bidding project, extracting structured evaluation indicators. Based on these indicators, it calculates the matching degree of historical bid documents, accurately identifying reference bid documents highly relevant to the current bidding project. Furthermore, this invention leverages the generative capabilities of the AI ​​model to generate multiple virtual bidding schemes based on the reference bid documents, providing the bidding party with multi-dimensional strategy predictions and optimization suggestions, significantly improving the intelligence level and decision-making reference value of bidding evaluation.

[0017] The further effects of the aforementioned non-conventional preferred method will be explained below in conjunction with specific embodiments. Attached Figure Description

[0018] To more clearly illustrate the embodiments of the present invention or the existing technical solutions, 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 recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram illustrating an application scenario of a bidding data evaluation method based on an AI large model, as provided in an embodiment of the present invention. Figure 2 A flowchart illustrating a bidding data evaluation method based on an AI large model, provided as an embodiment of the present invention; Figure 3 A flowchart illustrating another method for evaluating bidding data based on an AI large model, provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] The following will describe in detail, with reference to the accompanying drawings, a bidding data evaluation method based on an AI large model according to an embodiment of this application.

[0022] Figure 1 This is a schematic diagram illustrating an application scenario according to an embodiment of this application. The application scenario may include terminal devices 101, 102, and 103, server 104, and network 105.

[0023] Terminal devices 101, 102, and 103 can be hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays that support communication with server 104, including but not limited to smartphones, tablets, laptops, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. Terminal devices 101, 102, and 103 can be implemented as multiple software programs or software modules, or as a single software program or software module; this application embodiment does not impose any limitations on this. Furthermore, various applications can be installed on terminal devices 101, 102, and 103, such as data processing applications, instant messaging tools, social platform software, search applications, shopping applications, etc.

[0024] Server 104 can be a server that provides various services, such as a backend server that receives requests sent by terminal devices with which it has established communication connections. This backend server can receive and analyze the requests sent by the terminal devices and generate processing results. Server 104 can be a single server, a server cluster consisting of several servers, or a cloud computing service center. This application embodiment does not limit this.

[0025] It should be noted that server 104 can be either hardware or software. When server 104 is hardware, it can be various electronic devices that provide various services to terminal devices 101, 102, and 103. When server 104 is software, it can be multiple software programs or software modules that provide various services to terminal devices 101, 102, and 103, or it can be a single software program or software module that provides various services to terminal devices 101, 102, and 103. This application embodiment does not impose any limitations on this.

[0026] Network 105 can be a wired network using coaxial cable, twisted pair, and fiber optic connection, or it can be a wireless network that enables interconnection of various communication devices without wiring, such as Bluetooth, Near Field Communication (NFC), and Infrared. This application embodiment does not limit this.

[0027] Users can establish a communication connection with server 104 via network 105 through terminal devices 101, 102, and 103 to receive or send information, etc. Specifically, in response to an evaluation instruction for a target bidding project, server 104 obtains the tender documents for the target bidding project and extracts multiple historical tender documents associated with the target bidding project from the historical database of the electronic bidding platform; it inputs the tender documents and multiple historical tender documents into a pre-built AI big data model, which parses the tender documents and extracts evaluation indicators for at least one evaluation dimension; based on the evaluation indicators, the AI ​​big data model calculates the matching degree of each historical tender document and generates a matching degree score for each historical tender document under the evaluation indicators; based on the matching degree score, server 104 selects at least one reference tender document from the multiple historical tender documents and determines at least one virtual bidding scheme based on the reference tender document and the evaluation indicators using the AI ​​big data model; based on the matching degree score and the virtual bidding scheme, it generates an evaluation report for the target bidding project using the AI ​​big data model.

[0028] It should be noted that the specific types, quantities, and combinations of terminal devices 101, 102, and 103, server 104, and network 105 can be adjusted according to the actual needs of the application scenario, and this application embodiment does not impose any restrictions on this.

[0029] Figure 2 This invention provides a specific embodiment of a bidding data evaluation method based on an AI large-scale model. It is applied to an electronic bidding platform. This electronic bidding platform is an online bidding management system based on a B / S architecture, possessing functional modules such as bidding information release, bid document upload, expert review, and result announcement, and maintaining a historical database containing all historical bidding data. The following detailed description of the method in this embodiment, along with specific steps, further illustrates its implementation.

[0030] Step 201: In response to the evaluation instruction for the target tender project, obtain the tender documents for the target tender project and extract multiple historical tender documents associated with the target tender project from the historical database of the electronic tendering platform; Specifically, the bidding party's personnel select a bidding project with a published bidding announcement as the target project on the electronic bidding platform's interface and trigger an evaluation instruction. This evaluation instruction contains a unique identifier for the target project. In response to this instruction, the platform retrieves the tender document for the target project from its file server based on the unique identifier. This tender document is a structured or semi-structured electronic document containing a project overview, technical specifications, commercial terms, and bidder qualification requirements. Simultaneously, the platform extracts multiple historical bids associated with the target project from its historical database. To achieve precise correlation, the platform first obtains the project type and scale of the target project. Project type refers to the classification of projects according to national economic industry classifications or bidding practices, such as "municipal road engineering," "intelligent transportation system procurement," and "information technology software development services." Project scale is characterized by quantitative indicators such as project budget, workload, and estimated contract price, for example, a budget of 5 million yuan. Subsequently, based on the project type and scale, the platform performs a filtering operation from the historical database, extracting historical bids with the same project type and project scale within a preset similarity range. The preset similarity range can be set to fluctuate within 20% of the project size. For example, for a project with a budget of 5 million yuan, historical bid documents with a budget between 4 million and 6 million yuan can be extracted. This screening strategy ensures that the extracted historical bid documents are comparable to the target bidding project in terms of project nature and scale, thus providing an effective reference sample for subsequent evaluation.

[0031] Step 202: Input the tender documents and multiple historical tender documents into the pre-built AI big model. The AI ​​big model parses the tender documents and extracts evaluation indicators for at least one evaluation dimension. Furthermore, the platform inputs the tender documents obtained in step 201, along with multiple extracted historical tender documents, into a pre-built AI model. This AI model is a pre-trained large language model, whose underlying architecture can employ Transformer, BERT, or their derivatives. This large language model is not a general model trained directly on a general corpus; instead, it undergoes domain-fine-tuning using historical tender text corpora accumulated within the electronic tendering platform. Domain-fine-tuning refers to retraining the pre-trained model using specialized texts in the tendering field, including tender documents, bid documents, review reports, and winning bid announcements, enabling the model to accurately understand tendering-specific terms and semantic relationships such as "technical deviation table," "commercial quotation," "qualification certificate," and "performance certificate."

[0032] The process of the AI-powered big data model parsing the tender documents includes the following steps: First, the AI ​​model utilizes its built-in natural language processing module to perform text segmentation and semantic recognition on the tender documents. Text segmentation involves dividing the tender documents into text units at the granular level, such as chapters, paragraphs, and sentences, to form easily processed text units. Semantic recognition involves performing lexical analysis, syntactic analysis, and semantic understanding on the segmented text units to identify the key requirement fields contained within. Key requirement fields are specific contents explicitly stated in the tender documents that have binding or guiding requirements for bidders, such as "the project must be delivered within 90 days after the contract is signed," "the system must pass the Level 3 Information Security Protection Certification," and "the bidder must provide at least two similar project performance records within the past three years." Subsequently, the AI ​​model maps the identified key requirement fields to a pre-defined evaluation indicator system, generating structured evaluation indicators. The pre-defined evaluation indicator system is a pre-constructed classification framework used to standardize evaluation dimensions and indicators.

[0033] In this embodiment, the evaluation dimensions include at least the technical solution dimension, the commercial quotation dimension, and the enterprise qualification dimension. Accordingly, the generated evaluation indicators include: key technical points corresponding to the technical solution dimension, such as specific technical requirements like "system response time," "data interface standards," and "security protection level"; quotation ranges corresponding to the commercial quotation dimension, such as suggested quotation ranges based on project budget and market price analysis; and qualification level requirements corresponding to the enterprise qualification dimension, such as mandatory or rating-based qualification requirements like "Level 1 Qualification for Electronic and Intelligent Engineering Professional Contracting" and "ISO9001 Quality Management System Certification."

[0034] Step 203: Using the AI ​​big data model, based on the evaluation indicators, calculate the matching degree of each historical bid document and generate a matching degree score for each historical bid document under the evaluation indicators. Furthermore, after extracting the evaluation metrics, the AI ​​big data model performs a quantitative evaluation of each historical bid document based on these metrics. Specifically, the AI ​​big data model calculates the matching score of each historical bid document under the three evaluation dimensions mentioned above.

[0035] For the technical solution dimension, the AI ​​model iterates through each key technical point extracted in step 202, checking whether the technical solution section of the historical bid documents responds to, covers, or satisfies that key technical point. The model can use methods such as keyword matching, semantic similarity calculation, or deep learning classification to assign a compliance score (e.g., a value between 0 and 1) to each key technical point, and then combine the compliance scores of all key technical points to generate the technical matching degree of the historical bid document.

[0036] For the commercial pricing dimension, the AI ​​model extracts pricing amounts from historical bid documents and compares them with the pricing range determined in step 202. If the price falls within the range, the pricing match is considered high; if the price is below the lower limit or above the upper limit, the pricing match is reduced accordingly based on the degree of deviation. The degree of deviation can be calculated as a percentage of deviation, and the matching degree is negatively correlated with the degree of deviation.

[0037] Regarding the enterprise qualification dimension, the AI ​​model extracts enterprise qualification certificate information provided in historical bidding documents and compares it with the qualification level requirements determined in step 202. If the bidder's qualifications meet or exceed the requirements, the qualification matching degree is full or high; if the requirements are not met, the qualification matching degree is zero or extremely low.

[0038] After obtaining the matching scores for technology, pricing, and qualifications, the AI ​​model performs a weighted summation based on preset weights to determine the overall matching score of the historical bid document. The preset weights can be set based on bidding review practices, such as 50% for the technical solution dimension, 30% for the commercial pricing dimension, and 20% for the enterprise qualifications dimension. The weights can also be adjusted by the user according to the specific characteristics of the target bidding project; for example, for technology-intensive projects, the weight of the technical solution dimension can be increased.

[0039] Step 204: Based on the matching score, select at least one reference bid document from multiple historical bid documents, and determine at least one virtual bid scheme based on the reference bid document and evaluation indicators using an AI big data model. Further, the AI ​​big data model sorts the multiple historical bid documents in descending order based on the matching score calculated in step 203, and selects several top-ranked historical bid documents as reference bid documents. Reference bid documents refer to bid documents that have performed well in historical bidding activities and have a high degree of similarity and reference value to the current target bidding project. Subsequently, the AI ​​big data model generates at least one virtual bidding scheme based on the selected reference bid documents and the evaluation indicators extracted in step 202. This generation process specifically includes: inputting the reference bid documents as generation samples into the AI ​​big data model; and using the generative module of the AI ​​big data model (e.g., a Transformer-based decoder structure), combined with the evaluation indicators, performing parameter mutation and strategy optimization based on the reference bid documents. Parameter mutation refers to adjusting key parameters in the reference bid documents, such as fluctuating the bid amount by a certain percentage above or below the reference bid, or replacing a certain type of hardware configuration in the technical solution with an alternative solution with similar performance but lower cost. Strategy optimization refers to integrating the advantages of multiple reference bids. For example, combining Company A's core technical architecture with Company B's cost control strategy can generate a new solution that achieves a better balance between technological advancement and economic efficiency. Through this method, the AI ​​big data model can generate one or more virtual bid proposals that differ from any original reference bid in terms of technical solutions and commercial pricing. These virtual bid proposals are not submitted by real bidders, but rather are strategically competitive solutions created by the AI ​​big data model based on historical data and current tender requirements, providing the tendering party with decision-making references that transcend historical experience.

[0040] Step 205: Based on the matching score and the virtual bidding scheme, use the AI ​​big data model to generate an evaluation report for the target bidding project.

[0041] Furthermore, after completing the screening of reference bid documents and the generation of virtual bidding schemes, the AI ​​big data model integrates the above data to generate an evaluation report for the bidding party. Specifically, the AI ​​big data model uses the reference bid documents screened in step 204 and their corresponding matching scores, as well as the virtual bidding schemes generated in step 204, as input data. Combined with the evaluation indicators extracted in step 202, it uses its generative module to output a structured evaluation report.

[0042] The matching score plays multiple roles in this process: First, as a quantitative evaluation criterion for reference bid documents, the matching score is directly incorporated into the "Reference Case Comparison Analysis" section of the evaluation report. It visually displays the degree of fit between each historical bid document and the current target bidding project in a score-ranked manner, helping the bidding party quickly identify the most valuable cases. Second, when generating predictions of the winning probability of virtual bidding schemes, the AI ​​big data model uses the matching score of the reference bid documents as a benchmark, combined with the parameter variation of the virtual bidding scheme relative to the reference bid documents, to calculate the winning probability of the virtual scheme through simulation. For example, if a reference bid document has a matching score of 85 points and its historical winning record is successful, and the AI ​​big data model generates a virtual bidding scheme based on this reference document that optimizes the technical solution and lowers the price, the model can predict that the winning probability of the virtual scheme increases to 92%. Third, the matching score is also used to generate comprehensive evaluation conclusions. The AI ​​big data model extracts the common features of highly matched reference bid documents as "advantage features" and integrates them into pricing strategy suggestions and technical solution optimization suggestions, thereby providing the bidding party with data-supported decision-making references.

[0043] The evaluation report includes the following: a comparative analysis table of reference bid documents, listing the bidder's name, technical matching degree, price matching degree, qualification matching degree, and overall matching degree score for each reference bid document; a prediction of the probability of winning the bid, which is a quantitative estimate of the likelihood of winning the bid calculated based on historical winning patterns and matching degree scores for each reference bid document and virtual bidding scheme; pricing strategy recommendations, which recommend the optimal bid ceiling range or budget allocation scheme to the tendering party based on the price matching degree analysis; technical solution optimization suggestions, which extract common features of technical solutions in highly matching reference bid documents and propose directions for setting technical requirements that may obtain higher scores in the tender review; and potential risk warnings, which identify common problems in reference bid documents or performance risks hidden in virtual bidding schemes and provide corresponding risk avoidance suggestions. The evaluation report is presented in text form, supplemented by visual elements such as score comparison charts and radar charts, for the tendering party's personnel to review and use as a reference for decision-making.

[0044] As can be seen from the above technical solutions, the beneficial effects of this embodiment are: by combining the matching degree score with the virtual bidding scheme generated by the AI ​​large model, intelligent evaluation and strategic prediction of bidding data are realized, which significantly improves the accuracy and efficiency of the bidding party's decision-making.

[0045] Figure 2 The embodiments shown are merely basic examples of the method of the present invention. Other preferred embodiments of the method can be obtained by making certain optimizations and extensions based on them.

[0046] like Figure 3The image shows another specific embodiment of the bidding data evaluation method based on an AI large model according to the present invention. This embodiment further describes the method based on the foregoing embodiments, and includes the following steps:

[0047] Step 301: In response to user feedback on the electronic bidding platform, obtain user feedback data on the evaluation report; Specifically, this embodiment, based on the above embodiments, further introduces a feedback optimization mechanism to achieve continuous iteration and performance improvement of the AI ​​large model through user interaction. This mechanism enables the AI ​​large model to continuously adapt to the dynamic changes in user preferences and actual bidding scenarios, forming a data-driven closed-loop optimization system. Specifically, after the evaluation report generated in the above embodiments is presented to the bidding party, the electronic bidding platform provides an interactive interface allowing users to evaluate, correct, or supplement the content of the evaluation report. This user is typically a project manager, review expert, or decision-maker responsible for bidding work on the bidding party, possessing in-depth understanding of the background, technical requirements, and business conditions of the target bidding project. Feedback operations include, but are not limited to: approving or rejecting the selection results of the reference bid documents in the evaluation report; manually adjusting the technical parameters or quotation amount in the virtual bidding scheme; scoring the confidence level of the winning probability prediction results; adopting or rejecting the bidding strategy suggestions and technical solution optimization suggestions; and submitting supplementary opinions or corrections through text input boxes.

[0048] The platform responds to user feedback by acquiring and recording feedback data in real time. This feedback data is a structured or semi-structured collection of data, specifically including: a user identifier to distinguish the source of feedback from different users; a feedback timestamp to record the time the feedback occurred; a feedback type identifier to mark the type of feedback action (e.g., "adopt suggestion," "modify parameters," "text notes," etc.); feedback content data, i.e., the specific parameter values ​​modified, options selected, or text content entered by the user; and associated information, i.e., the evaluation report identifier, the target bidding project identifier, and the index information of the relevant reference bid documents or virtual bidding schemes. The platform associates and stores this feedback data with the original evaluation report, the input historical bid documents, and the intermediate processing results of the AI ​​large model, forming a complete data link that can be used for subsequent training.

[0049] Step 302: Use the feedback data as fine-tuning training data to optimize the parameters of the large AI model and update the large AI model.

[0050] Furthermore, the platform uses the feedback data obtained in step 301 as fine-tuning training data to optimize the parameters of the large AI model. The fine-tuning training data is a set of training samples suitable for supervised learning or reinforcement learning modes, formed by preprocessing and formatting the feedback data. Specifically, the platform first cleans and labels the feedback data: for suggestions adopted by the user, the output of the corresponding large AI model is labeled as a positive sample; for suggestions rejected or modified by the user, the correct result after modification is used as the target output, and the original generated result of the large AI model is used as a negative sample or a sample to be corrected for comparison and labeling; for text annotations added by the user, key correction information is extracted using natural language processing technology and transformed into structured labeled data.

[0051] After data labeling, the platform optimizes the parameters of the large AI model using the labeled, fine-tuned training data. This parameter optimization employs algorithms such as Supervised Fine-Tuning (SFT) or Reinforcement Learning from Human Feedback (RLHF). In Supervised Fine-Tuning mode, the platform uses the input portion of the labeled data (i.e., the original evaluation context, including tender documents, historical bid documents, evaluation metrics, etc.) as model input and the target output portion (i.e., the correct result confirmed or modified by the user) as the supervision signal. The platform adjusts the weight parameters of the large AI model through backpropagation, enabling the model to generate outputs closer to user expectations under similar input conditions. In Reinforcement Learning mode, the platform transforms user feedback data into reward signals, rewarding or penalizing the generated results of the large AI model. The platform optimizes the model's generation strategy through policy gradient algorithms, allowing the model to gradually learn output patterns that conform to user preferences. After parameter optimization, the platform deploys the updated large AI model online, replacing the original model version, for subsequent bidding data evaluation tasks. Through this feedback optimization mechanism, the AI ​​model can continuously absorb users' professional knowledge and decision-making preferences. The resulting evaluation reports and virtual bidding proposals will increasingly align with actual business needs, leading to continuous improvement in evaluation accuracy and user satisfaction. Simultaneously, this mechanism also accumulates high-quality domain-specific labeled data for the electronic bidding platform, laying a data foundation for the model's long-term evolution.

[0052] As can be seen from the above technical solutions, the beneficial effects of this embodiment are: by introducing a closed-loop optimization mechanism that combines user feedback and model fine-tuning, the AI ​​large model can achieve continuous learning and personalized adaptation in bidding and tendering evaluation scenarios, which significantly improves the intelligence level and practicality of the evaluation system.

[0053] Figure 4This is a schematic diagram of the electronic device 4 provided in an embodiment of this application. Figure 4 As shown, the electronic device 4 of this embodiment includes a processor 401, a memory 402, and a computer program 403 stored in the memory 402 and executable on the processor 401. When the processor 401 executes the computer program 403, it implements the steps in the various method embodiments described above.

[0054] Electronic device 4 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 4 may include, but is not limited to, processor 401 and memory 402. Those skilled in the art will understand that... Figure 4 This is merely an example of electronic device 4 and does not constitute a limitation on electronic device 4. It may include more or fewer components than shown, or different components.

[0055] The processor 401 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0056] The memory 402 can be an internal storage unit of the electronic device 4, such as a hard disk or RAM of the electronic device 4. The memory 402 can also be an external storage device of the electronic device 4, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the electronic device 4. The memory 402 can also include both internal and external storage units of the electronic device 4. The memory 402 is used to store computer programs and other programs and data required by the electronic device.

[0057] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0058] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a readable storage medium (e.g., a computer-readable storage medium). Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which may be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable storage medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0059] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for evaluating bidding data based on an AI large-scale model, characterized in that, Applied to electronic bidding platforms, the method includes: In response to an evaluation instruction for a target tender project, the tender documents for the target tender project are obtained, and multiple historical tender documents associated with the target tender project are extracted from the historical database of the electronic tendering platform. The tender document and the multiple historical tender documents are input into a pre-built AI big model. The AI ​​big model parses the tender document and extracts evaluation indicators for at least one evaluation dimension. Using the AI ​​big data model and based on the evaluation indicators, the matching degree of each historical bid document is calculated, and a matching degree score of each historical bid document under the evaluation indicators is generated. Based on the matching score, at least one reference bid document is selected from the multiple historical bid documents, and at least one virtual bid scheme is determined by the AI ​​big data model based on the reference bid document and the evaluation index. Based on the matching score and the virtual bidding scheme, the AI ​​big data model is used to generate an evaluation report for the target bidding project.

2. The method according to claim 1, characterized in that, The evaluation dimensions include at least the technical solution dimension, the business quotation dimension, and the enterprise qualification dimension; The evaluation indicators include key technical points corresponding to the technical solution dimension, price range corresponding to the business quotation dimension, and qualification level requirements corresponding to the enterprise qualification dimension.

3. The method according to claim 2, characterized in that, Using the AI ​​big data model and based on the evaluation metrics, a matching degree calculation is performed on each of the historical bid documents to generate a matching degree score for each historical bid document under the evaluation metrics, including: Using the AI ​​big model, the technical matching degree of each of the historical tender documents is calculated under the technical solution dimension, the price matching degree under the business pricing dimension, and the qualification matching degree under the enterprise qualification dimension. Based on a preset weight allocation, the matching degree of technology, the matching degree of price, and the matching degree of qualifications are weighted and summed to determine the matching degree score.

4. The method according to claim 1, characterized in that, The step of obtaining the tender documents for the target tender project and extracting multiple historical tender documents associated with the target tender project from the historical database of the electronic bidding platform includes: Obtain the project type and project scale of the target bidding project; Based on the project type and the project size, extract historical bid documents from the historical database that have the same project type and a project size within a preset similarity range.

5. The method according to claim 1, characterized in that, The process involves parsing the tender document using the AI ​​model to extract evaluation metrics for at least one evaluation dimension, including: The natural language processing module in the AI ​​model is used to perform text segmentation and semantic recognition on the tender document to identify the key requirement fields in the tender document; The key requirement fields are mapped to a preset evaluation indicator system to generate the structured evaluation indicators.

6. The method according to claim 1, characterized in that, The process of generating at least one virtual bidding scheme based on the reference bid documents and the evaluation indicators using the AI ​​big data model includes: The reference tender document is used as a sample to be input into the AI ​​large model; Using the generative module of the AI ​​big model and the evaluation indicators, parameter mutations and strategy optimizations are performed on the reference bid documents to generate one or more virtual bid proposals that differ from the reference bid documents in terms of technical solutions and commercial quotations.

7. The method according to claim 1, characterized in that, The step of generating an evaluation report for the target bidding project using the AI ​​big data model based on the matching score and the virtual bidding scheme includes: The selected reference bid documents and their corresponding matching scores, along with the generated virtual bidding scheme, are used as comparative analysis data and input into the AI ​​big model. The generative module of the AI ​​big data model, combined with the evaluation indicators, generates an evaluation report that includes predictions of the probability of winning the bid, suggestions on pricing strategies, suggestions for optimizing technical solutions, and warnings of potential risks.

8. The method according to claim 1, characterized in that, Also includes: In response to user feedback on the electronic bidding platform, obtain user feedback data on the evaluation report; The feedback data is used as fine-tuning training data to optimize the parameters of the large AI model, thereby updating the large AI model.

9. The method according to claim 1, characterized in that, The AI ​​big model is a pre-trained big language model, and the big language model has been fine-tuned in the domain using historical bidding text corpus in the electronic bidding platform.

Images