An automated bid evaluation method, apparatus, equipment and system for bidding projects.

By using automated evaluation methods, combined with least squares calculations and historical data analysis, and dynamically adjusting evaluation weights, the problem of traditional evaluation relying on manual judgment is solved. This achieves standardization and reliability of evaluation results, and improves evaluation efficiency and accuracy.

CN123117909APending Publication Date: 2026-11-17FUAN QINGMEI ENERGY MATERIALS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611335241.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-31
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Traditional bid evaluation processes rely on manual judgment, resulting in inconsistent and unreliable evaluation results. They also lack quantitative indicators and cannot adapt to the differences in the nature of bids submitted by various bidders.

Method used

An automated evaluation method is adopted, which uses least squares calculation combined with historical bidding data and bidding behavior analysis to dynamically adjust weights and generate comprehensive evaluation results, reducing human intervention.

Benefits of technology

It achieves standardization and reliability of bid evaluation results, reduces manual operations, improves bid evaluation efficiency and accuracy, and adapts to the bid evaluation needs of different bidders.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117909A_ABST
    Figure CN123117909A_ABST
Patent Text Reader

Abstract

This invention provides an automated bid evaluation method, apparatus, equipment, and system for bidding projects, relating to the field of intelligent review technology. The method includes: determining the subjective weight of the bidding party on preset scoring indicators in the bidding project documents and the objective weight of the bidders on preset evaluation indicators, thus obtaining a target comprehensive weight; retrieving historical bidding documents based on project features extracted from the project content in the bidding project documents, and interpolating the target comprehensive weight based on its historical comprehensive weight to obtain a target dynamic weight; and correcting the target comprehensive weight based on the risk weight of the bidders' bidding risks to obtain a final weight, which is then used to weight the initial evaluation score of the bidders' bid documents after bid review. Based on the weighted comprehensive score, the bidders' evaluation results for the bidding project documents are generated. This invention addresses the problem that existing technologies still rely on manual judgment and have rigid evaluation methods for reviewing bidding documents, leading to low reliability of evaluation results.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of intelligent review technology, specifically to an automated bid evaluation method, apparatus, equipment, and system for bidding projects. Background Technology

[0002] Manufacturing projects (such as factory construction, production line renovation, and workshop decoration) require a large number of small-scale engineering projects each year. These projects are characterized by their large number, small amount, and scattered timeframes. Traditional bidding processes typically employ paper-based bids, manually formed bid evaluation committees, and on-site bid evaluations. However, these methods rely heavily on manual labor, consuming significant time and severely disrupting normal production management.

[0003] While existing technologies propose using electronic bidding systems for electronic review and submission of bid documents online, they have several drawbacks. Firstly, various evaluation indicators still rely on manual judgment and lack quantifiable metrics, making evaluation results susceptible to influence from personal experience and preferences, leading to inconsistent evaluation standards. Secondly, evaluation methods, such as fixed and rigid weighting, fail to consider the differences in the nature of each bidder's bid, resulting in low reliability of the evaluation results. Summary of the Invention

[0004] In view of this, it is necessary to provide an automated bid evaluation method, device, equipment and system for bidding projects, in order to solve the problem that the existing technology still relies on manual judgment for the review of bidding documents and the bid evaluation methods are fixed and rigid, resulting in low reliability of bid evaluation results.

[0005] To address the above problems, this invention provides an automated bid evaluation method for bidding projects, comprising: Obtain the bidding project documents from the bidding party and the initial evaluation scores of the bid documents after bid review from the bidders; Determine the subjective weight of the tendering party on the preset scoring indicators corresponding to the tender project documents, determine the objective weight of the bidder on the preset evaluation indicators, and perform least squares calculation on the subjective weight and the objective weight to obtain the target comprehensive weight. Extract project features from the project content in the bidding project documents, retrieve historical bidding documents from a preset historical bidding project database based on the project features, and interpolate the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight; The risk weight for the bidder's bidding risk is determined based on the bidder's historical bidding behavior data, and the target dynamic weight is adjusted based on the risk weight to obtain the final weight. The initial evaluation score is weighted according to the final weight, and the bid evaluation result of the bidder on the bidding project documents is generated based on the weighted comprehensive score.

[0006] In one possible implementation, the tender documents are obtained through the following method: Obtain the drawings and documents for the bidding project; The drawing file is parsed for content elements, and the parsed file is then checked for content integrity. The quantities of the project content in the file after integrity verification are calculated, and the isolated forest model is called to detect anomalies in the calculated quantities. When the anomaly detection result is no anomaly, a cost database for the project content is constructed; The cost database, the project quantity, and the integrity-verified file are bound together to obtain the bidding project document.

[0007] In one possible implementation, the preset scoring indicators include: credit indicators, technical indicators, content indicators, price indicators, and balance indicators. The initial evaluation score is obtained by calculating the score of the tender document on the preset scoring indicators as the initial evaluation score. The calculation of the scores of the bid documents on the preset scoring indicators includes: The tender documents are subject to qualification review to obtain a qualification score on the creditworthiness indicator. The tender document is segmented into words, and the segmented content is fuzzy matched based on a preset terminology database to obtain a technical standardization score in terms of technical indicators. The semantic similarity of the content of the tender documents is calculated to obtain a bid-rigging risk score on the content indicator. Calculate the matching degree between the price value and the corresponding budget value of each item in the tender document to obtain the price deviation score on the price indicator; The TOPSIS multi-criteria decision model is used to calculate the relative closeness between the bid prices recorded in each tender document and the benchmark price, and the bid imbalance score on the balance index is obtained.

[0008] In one possible implementation, the step of performing a least-squares calculation on the subjective weights and the objective weights to obtain the target comprehensive weights includes: Construct a least squares model, wherein the objective function of the least squares model is the sum of squares of a first difference and a second difference, the first difference being the difference between the target comprehensive weight and the subjective weight, and the second difference being the difference between the target comprehensive weight and the objective weight; The least squares model is optimized with minimizing the sum of squares as the optimization objective to obtain the target comprehensive weight.

[0009] In one possible implementation, after extracting the project features from the project content in the bidding project document, the method further includes: Obtain a preset weight adjustment rule for each feature item of the project features, wherein the preset weight adjustment rule records the comprehensive weight adjustment value for each feature item; Each feature item in the project features is judged according to the preset weight adjustment rule; When a feature item meets the preset weight adjustment rules, obtain the comprehensive weight adjustment value of the corresponding feature item; A correction coefficient vector is generated based on the comprehensive weight adjustment value, and the target comprehensive weight is adjusted based on the correction coefficient vector.

[0010] In one possible implementation, the step of interpolating the target comprehensive weight based on the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight includes: Obtain the similarity value when retrieving the historical tender documents based on the project features; The similarity value is determined as the interpolation coefficient for interpolation calculation; The target dynamic weight is obtained by interpolating the historical comprehensive weight of the historical bidding documents and the target comprehensive weight based on the interpolation coefficient.

[0011] In one possible implementation, determining the risk weight for bidding risks based on the bidder's historical bidding behavior data includes: Bidding behavior identification is performed on the historical bidding behavior data of each bidder, resulting in multiple behavior types, including abnormal behavior of non-performance, performance behavior, and collusion behavior. For each type of behavior, a bidding risk assessment is conducted separately to obtain a risk level score; The risk level score is weighted based on a preset penalty coefficient, and the weighted results are summed to obtain the risk weight of the bidder's bidding risk.

[0012] The present invention also provides an automated bid evaluation device for bidding projects, comprising: The acquisition module is used to acquire the bidding project documents from the bidding party and the initial evaluation scores of the bidders' bid documents after review. The determination module is used to determine the subjective weight of the tendering party on the corresponding scoring indicators in the tender project document, determine the objective weight of the bidder on the preset evaluation indicators, and perform least squares calculation on the subjective weight and the objective weight to obtain the target comprehensive weight. The calculation module is used to extract project features from the project content in the bidding project file, retrieve historical bidding documents from a preset historical bidding project database based on the project features, and interpolate the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight. The correction module is used to determine the risk weight of the bidder's bidding risk based on the bidder's historical bidding behavior data, and to correct the target dynamic weight based on the risk weight to obtain the final weight; The bid evaluation module is used to weight the initial evaluation score according to the final weight, and generate the bid evaluation result of the bidder on the bidding project documents based on the weighted comprehensive score.

[0013] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a program; the processor is coupled to the memory and is used to execute the program stored in the memory to implement the steps of the above-described automated bid evaluation method for bidding projects.

[0014] The present invention also provides an automated bid evaluation system for bidding projects, including a data interface device, a communication device, and the aforementioned electronic device. The data interface device communicates with the electronic device through the communication device. The data interface device is used to receive and acquire the bidding project documents of the bidding party and the bid documents of the bidding party. The communication device is used to transmit the bidding project documents and the bid documents to the electronic device, and after the processor of the electronic device executes the program stored in the memory, it transmits the bid evaluation results of the output bidding project documents to the data interface device.

[0015] The beneficial effects of adopting the above implementation method are as follows: The automated bid evaluation method for bidding projects provided by this invention first evaluates the bidding documents separately on subjective and objective weights, and then synthesizes them using least squares. This preserves the bidding party's priority judgment on scoring indicators such as price and technology, while also accurately reflecting the objective differences in the bid document data. Furthermore, by extracting project characteristics and retrieving similar historical bidding documents, historical bidding experience is used to assist in the review and adjust the target comprehensive weights without relying on manual judgment, making the evaluation weight configuration results more credible. In addition, by analyzing the bidders' historical bidding behavior to determine the risk weights of bidding risks, a "penalty" is imposed on bidders, thereby correcting the weight configuration and eliminating the influence of multiple bidders with varying bid characteristics. This ensures that the final weights accurately measure the bidders' overall performance, making the evaluation results more standardized and uniform. Moreover, the entire evaluation process can be automated, without any manual operation or reliance on any manual judgment, effectively improving the reliability of the evaluation results. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the automated bid evaluation method for bidding projects provided by this invention; Figure 2 This is a schematic diagram of the structure of the automated bid evaluation device for bidding projects provided by the present invention; Figure 3 A schematic diagram of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0019] In the description of the embodiments of this application, unless otherwise stated, "a plurality of" means two or more.

[0020] In this embodiment of the invention, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, apparatus, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such process, method, product or device.

[0021] The naming or numbering of steps in the embodiments of the present invention does not mean that the steps in the method flow must be executed in the time / logical order indicated by the naming or numbering. The execution order of the named or numbered process steps can be changed according to the technical purpose to be achieved, as long as the same or similar technical effect can be achieved.

[0022] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] The automated bid evaluation method for bidding projects provided in this invention can be applied to bid evaluation scenarios for various project types. The executing entity can be a server, terminal, or remote cloud device. By collecting and uploading the bidding project documents from the bidding party and the bid documents from the bidders, and then calling the automated bid evaluation method provided by this invention, automated bid evaluation is achieved, and the final bid evaluation results of the bidding project documents are output, providing the bidding party with a reliable theoretical basis for bid evaluation.

[0024] The following section details the automated bid evaluation method for bidding projects provided by this invention.

[0025] Figure 1 This is a flowchart illustrating the automated bid evaluation method for bidding projects provided by the present invention, as shown below. Figure 1 As shown, the automated bid evaluation method for bidding projects can be implemented through the following steps 101 to 105, which are explained in detail below.

[0026] Step 101: Obtain the bidding project documents from the bidding party and the initial evaluation scores of the bid documents after bid review from the bidders.

[0027] When evaluating bids for a tender project, the tendering party needs to provide a standardized and complete tender project document. In addition, bidders also need to provide compliant and bid-compliant bid documents for the tendering party's reference. In this embodiment of the invention, there are multiple bidders, and each bid document corresponds one-to-one with a bidder; that is, each bidder provides only one bid document. Typically, the tendering party provides the corresponding drawings for the tender project, not the standardized tender project document. Therefore, it is necessary to perform content parsing, content integrity verification, quantity calculation, and construction of a cost database to ultimately generate a tender project document that meets the tender requirements.

[0028] In addition, it is necessary to perform word segmentation and stop word removal on the bidding project documents and tender documents respectively, then calculate the TF-IDF vectors of the two documents, use cosine similarity to judge the responsiveness of each review clause (qualification conditions, technical standards, business requirements) in the two documents, and use regular expression matching to supplement the hard requirements such as providing a safety production license.

[0029] In one possible implementation, the tender documents can be obtained in the following ways, as detailed below.

[0030] Obtain the drawings and documents for the bidding project; The drawing file is parsed for content elements, and the parsed file is then checked for content integrity. The quantities of the project content in the file after integrity verification are calculated, and the isolated forest model is called to detect anomalies in the calculated quantities. When the anomaly detection result is no anomaly, construct the cost database for the project content; By binding the cost database, the project quantity, and the files after integrity verification, the bidding project documents are obtained.

[0031] Here, parsing the content elements of drawing files mainly involves data extraction and coordinate unification. Drawing files can be DWG files, and the open-source library LibreDWG can be used to read entity objects (line segments, arcs, polylines, blocks, dimensions, etc.), layer information (name, color, linetype, on / off status), and block attributes (label, default value, visibility) from the DWG file. For multiple drawings (e.g., different floors or disciplines), the lower left corner reference point or the origin of the World Coordinate System (WCS) is identified, and all graphic elements are transformed to the same global coordinate system. If there are nested blocks or external references, their transformation matrices are recursively parsed and applied to achieve geometric data normalization.

[0032] Furthermore, the process involves mapping graphic elements to text. Based on a unified coordinate system, spatial proximity and topological relationships are used to establish associations between graphic elements and labeled text. The measurement endpoints of dimension lines are matched with the endpoints of nearby line segments / arcs, and the corresponding graphic elements are bound to complete dimension annotation. The target graphic element indicated by the leader line is bound to the labeled text, completing leader line annotation. For isolated text such as component names and elevation values, the nearest distance between its bounding box and each graphic element is calculated, and it is assigned to the nearest wall, column, door, or window component.

[0033] After content parsing is complete, the next step is to perform content integrity verification on the parsed file. First, the number of drawings is verified. All layouts in the DWG file are traversed, and the number of non-empty layouts is counted. This number is compared to the number of drawing pages specified in the project directory. If the number is insufficient, the names of missing drawings are output (extracted from the layout title or title block text), indicating a failure of the content integrity verification. Next, the compliance of the signature section, key tables, and layers is verified. For the signature section, the attribute fields in the title block (e.g., "TitleBlock") are located, and required fields such as "signatory," "reviewer," and "date" are checked for emptiness or placeholders. Otherwise, the signature is incomplete, indicating a failure of the content integrity verification. For key tables, regular expressions are used to match titles such as "Doors and Windows Schedule" and "Reinforcement Schedule" in the DWG file, and the presence of a grid structure formed by horizontal / vertical lines is verified to confirm the table's availability. If not, the content integrity verification fails. For layer compliance, the layer list in the DWG file is compared with the project's standard layer list (such as A-WALL, A-DOOR, etc.). If missing or redundant layers are detected, the content integrity check is deemed to have failed.

[0034] Drawings that fail the content integrity check will not be included in the evaluation. Therefore, they will not be used to build bidding project documents. They can be rolled back for manual review and modification until the content integrity check is successful.

[0035] For files that have passed integrity verification, further quantity calculations can be performed. First, component classification is performed, extracting the geometric features (area, aspect ratio, number of vertices, layer name, adjacency relationships, etc.) of each closed element region. A classification model is trained using decision trees or random forests to automatically identify major components such as walls, beams, columns, doors, and windows. Then, attribute parsing is performed, using a rule engine and regular expressions to read component numbers, dimensions, elevations, reinforcement descriptions, and other information from text objects and block attributes. For example, a predefined mapping table (“KL-1” → “Frame beam, section 300×500”) and the dimension regular expression “(\d+)\s*[xX×]\s*(\d+)” are used to achieve subdivision of categories and attribute assignment. Finally, quantity calculations are performed by calling the CGAL geometric calculation library, combining spatial topology analysis and preset deduction rules (such as deducting door and window openings from wall volume), to automatically calculate the volume, area, length, quantity, and other quantities of each component. After the calculation is completed, its rationality is checked. The isolated forest model is called to detect anomalies in the calculated quantities and calculate the deviation of the quantities for each item. If the deviation is small (e.g., the deviation does not exceed the preset deviation threshold), it is judged to be without anomalies. If the deviation is large (the deviation exceeds the preset deviation threshold), the item is removed and will not participate in this bidding, or manual review is triggered. Once the manual review confirms that there are no anomalies, it is judged to be without anomalies.

[0036] When the anomaly detection result is no anomaly, a cost database is constructed for the project content in the file. This is done using a relational database (such as MySQL or PostgreSQL). The relational database uses the item list code from the project content as the primary key, storing columns for price items such as project name, feature description, unit of measurement, comprehensive unit price, material cost, labor cost, and machinery cost, supporting high-frequency and precise queries. Furthermore, using preset matching rules, matching methods such as item list code mapping or keyword search are used to match projects in the drawing file with the price items corresponding to the same item list code or keywords in the database, thus establishing the cost database. If no match is found, the item list from the project content, along with the corresponding price item fields (item name, feature description, unit of measurement, comprehensive unit price, material cost, labor cost, and machinery cost), is added to the cost database, achieving dynamic incremental updates.

[0037] Finally, the quantities are marked in the file after integrity verification to achieve binding, and the cost database is bound to the file after integrity verification to achieve file binding of project cost. The file after binding is complete can be used as the bidding project document.

[0038] In this embodiment of the invention, based on the drawings and documents of the bidding project, the preprocessing operation of the documents is automatically performed. The bidding project documents are automatically generated without the need for manual statistics, which can significantly save the document processing time before the bid evaluation and improve the efficiency of the bid evaluation.

[0039] Furthermore, this embodiment of the invention also achieves automated bid review for the bidders' bid documents, obtaining an initial evaluation score, which serves as an important basis for bid evaluation.

[0040] In one possible implementation, the initial evaluation score is obtained in the following manner, as detailed below.

[0041] Specifically, the scores of the tender documents on the preset scoring indicators are calculated as the initial evaluation scores.

[0042] The preset scoring indicators here are the key indicators that the project focuses on during the bidding process. These indicators have five dimensions, namely, five preset scoring indicators: creditworthiness, technical, content, price, and balance. The automated bid review process needs to be carried out from these five indicator dimensions, which will be introduced one by one below.

[0043] In the dimension of credit indicators, qualification compliance review is carried out. The bidding documents are subject to qualification review to obtain a qualification score on the credit indicator. Here, the TesseractOCR engine interface can be invoked to identify key fields in qualification certificates, personnel certificates, and performance contracts in the bidding documents, and through structured queries and authoritative data sources, such as invoking the Tianyancha Open Platform API (Token authentication, HTTP GET), to obtain in batches data such as paid-in capital, number of employees, and key personnel of the bidding enterprises in the bidding documents for cross-verification of authenticity. In addition, the social credit data of the bidding enterprise is judged, and the credit status of the bidding enterprise can be roughly evaluated based on the identified qualification content and social credit data to determine the credit level, which is mapped to a qualification score. The higher the qualification score on the credit indicator, the higher the qualification and social credit of the bidding enterprise, and the higher the qualification compliance; otherwise, the lower it is.

[0044] In the dimension of credit indicators, term standardization detection is carried out. Word segmentation processing is performed on the bidding documents, and fuzzy matching is performed on the segmented content after word segmentation based on a preset term base to obtain a technical standardization score on the technical indicator.

[0045] After obtaining the bidding project documents, the embodiment of the present invention pre-constructs a related standard term database according to the project content in the documents, including the standard names, synonyms, forbidden words and the like of various terms. For example, when the project content is construction engineering, the term database records the standard names, synonyms, forbidden words and the like of terms related to the project engineering. For the bidding documents, a language model or the like is invoked for word segmentation, and then the segmented content after word segmentation is subjected to fuzzy matching with the term database, such as retrieval matching, calculation of Levenshtein Distance, and the like. A similarity value can be calculated through fuzzy matching, and the similarity value can be used as the technical standardization score on the technical indicator. The larger the similarity value, the more accurate the use of terms and the higher the technical standardization, and vice versa. In addition, if it is determined that the similarity value between the segmented content and the term in the term database is greater than a preset threshold (such as 85%) but the expressions are inconsistent, it can be judged as a non-standard expression, and the matched term can replace the term in the segmented content. For some common non-standard abbreviations, a mapping dictionary can be established and supplemented with regular expressions for fast scanning and positioning, such as "tong" and "concrete", "shanghun" and "ready-mixed concrete", etc.

[0046] On the content metrics dimension, the focus is on identifying similar content and potential collusion. Semantic similarity is calculated on the content of the bid documents to obtain a collusion risk score. In actual bidding scenarios, there are multiple bidders and multiple bid documents. It is necessary to determine the similarity between different bid documents to identify any collusion. The specific method is to directly calculate the semantic similarity (e.g., cosine similarity) between the content of the bid documents, filtering out formatted content common to every bid document (such as company profiles and legal clauses). When the semantic similarity exceeds a preset similarity threshold (e.g., 80%), it indicates potential collusion. These bid documents with potential collusion are flagged and investigated, and excluded from the bidding process. If the semantic similarity is not greater than the preset similarity threshold, no collusion is considered, and this semantic similarity score serves as the collusion risk score on the content metrics.

[0047] At the price level, abnormal pricing is detected. The tender document records the price values ​​for each item in the bill of quantities, along with corresponding budget values ​​recorded in the budget file; the bid price cannot exceed the budget. Here, the matching degree between the price values ​​of each item in the bill of quantities and their corresponding budget values ​​is calculated. The project names, feature descriptions, and other project characteristics in the tender document are matched with their corresponding names and feature descriptions in the budget file. This matching process can involve calculating TF-IDF vectors or cosine similarity. Then, the price values ​​for items such as the comprehensive unit price, labor costs, material costs, and machinery costs in the tender document are compared item by item with their corresponding budget values ​​in the budget file. The deviation rate is calculated, and abnormal price values ​​(deviation rates exceeding a preset deviation threshold) are identified. This deviation rate can be mapped to a price deviation score on the price indicator; a higher deviation rate results in a higher price deviation score, and vice versa.

[0048] For any identified abnormal price values, the corresponding items in the price list will not participate in this bidding process. There may be multiple abnormal price values; a K-value can be used to identify them. The Means algorithm groups similar list items and identifies significant price deviations. Based on item characteristics (structure type, floor height, region, etc.), it trains a model using gradient boosting trees (e.g., XGBoost or LightGBM) to predict the reasonable price range for each list item, marking prices exceeding that range. Through a rule engine and threshold detection, items exceeding the agreed deviation range are automatically highlighted in red and given tiered alerts (e.g., minor deviation, severe deviation). Pre-defined categories of reasons for the differences (material price differences, construction process differences, fee standard differences) are used to populate a template based on the rule-based judgment results, outputting a structured difference report.

[0049] On the balance index dimension, the focus is on identifying unbalanced pricing in each bid document. This is achieved by collecting historical bid documents for the corresponding project content, selecting the median price among the bids as the positive ideal solution, and then using the largest absolute deviation as the negative ideal solution. The TOPSIS multi-criteria decision model is then used to calculate the relative closeness between the bid price and the preset benchmark price in each bid document, quantifying the degree of price imbalance among the bid documents. Furthermore, the calculated relative closeness serves as the price imbalance score on the balance index.

[0050] Finally, the preset scoring indicators calculated for each preset scoring indicator dimension are normalized and standardized in terms of units, and then combined to form a decision matrix. , recorded as Where i represents the i-th bid document, and j represents the j-th preset scoring indicator. This refers to the score corresponding to the j-th preset scoring indicator in the i-th bid document, where m and n represent the width and height of the decision matrix, respectively. The score values ​​recorded in this decision matrix are the initial evaluation scores of the bidder's bid documents after the bid review.

[0051] This invention, through five preset scoring indicators, automates the review of bid documents from five dimensions. Traditional qualitative factors such as budget rationality, technical compliance, and collusion are quantified into calculable scores, serving as initial reference values ​​for the evaluation results and incorporated into the weighted scoring system. This makes the evaluation results more objective and traceable. Simultaneously, the calculated scores enable the automated review process to detect anomalies such as bid plagiarism, collusion, abnormal pricing, and unbalanced pricing, providing effective risk warnings for the bidding party.

[0052] Step 102: Determine the subjective weight of the tendering party on the corresponding preset scoring indicators in the tender project documents, determine the objective weight of the bidder on the preset evaluation indicators, and perform least squares calculation on the subjective and objective weights to obtain the target comprehensive weight.

[0053] After processing the bidding project documents and bid documents through the above steps, the automated bid evaluation process can begin. The bid evaluation process involves assessing each bid document, estimating the final weight of the bid document using a weighted scoring system, achieving an effective evaluation of the bidders, and finally generating the bid evaluation results corresponding to the bidders.

[0054] In this embodiment of the invention, estimating the final weight of a bid document using a weighted scoring system first requires constructing a basic weighting system. This process involves a combination of subjective and objective weighting. Subjectively, the subjective weight of the bidding party on the pre-defined scoring indicators corresponding to the bidding project documents is determined. This can be achieved using the Analytic Hierarchy Process (AHP), which compares the importance of each scoring dimension (such as price, technology, and creditworthiness) pairwise to construct a judgment matrix and calculate the subjective weights. Specifically, a hierarchical structure is first constructed, setting a target layer (comprehensive scoring) and a criterion layer (e.g., price reasonableness, technical solution, corporate creditworthiness, project schedule commitment, etc., with n elements). A judgment matrix is ​​then constructed through pairwise comparisons. ,in, The subjective vector weights represent the importance of criterion i relative to criterion j (values ​​range from 1 to 9 depending on importance). Represented as:

[0055]

[0056] Where T denotes the transpose of the matrix. This represents the subjective vector weight values ​​corresponding to n criteria. This is the judgment matrix.

[0057] Furthermore, calculate the consistency ratio CR = CI / RI, where CI = ), The largest eigenvalue of the judgment matrix is ​​used, and RI is the average random consistency index. The judgment matrix is ​​considered acceptable when CR < 0.1.

[0058] On an objective level, the objective weights of each bidder on the pre-set evaluation indicators are determined. Specifically, the scores of each bid document on each pre-set evaluation indicator are first calculated and then standardized. Here, there are M bid documents, corresponding to M bidders, and m pre-set evaluation indicators. The scores of each bid document on the pre-set evaluation indicators are... The constructed data matrix is ​​X= For positive indicators (the larger the value, the better), take... ,in, That is Standardized value and These represent the minimum and maximum scores for the j-th preset evaluation indicator in each bid document, respectively. For negative indicators (smaller values ​​are better), take... If all bidders have the same value under a certain preset evaluation index (i.e., ... If the preset evaluation index has no distinguishing effect on all bidders, the system will automatically remove the preset evaluation index (it will not participate in the subsequent entropy weight calculation, and its objective weight will be directly assigned to 0) in order to avoid the denominator being zero during the standardization process.

[0059] For the retained valid pre-defined evaluation indicators, we calculate the characteristic weight of the i-th bidder on the j-th pre-defined evaluation indicator, denoted as . Further calculate the entropy value of the j-th preset evaluation index. , represented as Therefore, the objective weight of the j-th preset evaluation index The calculation formula is as follows:

[0060] Where m represents the number of preset evaluation indicators, and k represents the k-th preset evaluation indicator. This represents the entropy value of the k-th preset evaluation index.

[0061] Finally, the weights at the subjective and objective levels are combined. Here, the least squares method is used to calculate the combined weights. The least squares method is used to calculate the subjective weights and the objective weights to obtain the target comprehensive weight.

[0062] In one possible implementation, the subjective weights and objective weights are calculated using least squares to obtain the target comprehensive weights. This can be achieved in the following ways, which will be explained in detail below.

[0063] Construct a least squares model, where the objective function of the least squares model is the sum of squares of the first difference and the second difference. The first difference is the difference between the objective comprehensive weight and the subjective weight, and the second difference is the difference between the objective comprehensive weight and the objective weight. The least squares model is optimized with the goal of minimizing the sum of squares, and the target comprehensive weight is obtained.

[0064] Here, the search for combined weights is achieved through a least squares model. This model seeks a comprehensive weight that minimizes the total deviation from both subjective and objective weights during the optimization process. Therefore, the objective function of the model is to minimize the sum of squares of the total deviations between the objective comprehensive weight and the subjective and objective weights, expressed as:

[0065]

[0066] Where m represents the number of preset evaluation indicators, and These represent the subjective and objective weights of the j-th preset evaluation index, respectively. This represents the target comprehensive weight of the j-th preset evaluation index, which is also the target that the least squares model needs to optimize and find.

[0067] The optimization problem of the least squares model is a convex quadratic programming problem with an analytical solution, which is obtained by applying the objective function. Find the partial derivatives and set them to 0, then combine... The constraints can be calculated. for:

[0068] in, , Both represent the subjective and objective weights of the j-th preset evaluation index.

[0069] Using the same method described above, the target comprehensive weight is calculated for each preset evaluation indicator one by one. The target comprehensive weight of m preset evaluation indicators is expressed as follows: .

[0070] Step 103: Extract the project features from the project content in the bidding project documents, retrieve historical bidding documents from the preset historical bidding project database based on the project features, and interpolate the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight.

[0071] Here, for the project content in the bidding project documents, the corresponding project feature F is extracted first. Taking construction projects as an example, project feature F includes project type (such as civil engineering, installation, or decoration), building area, number of floors, technical complexity level, and time urgency. Taking data collection projects as an example, project feature F can be the type of product to be collected, the quantity to be collected, the allowed collection time, etc.

[0072] The following explanation will use construction projects as an example. Based on project characteristics F, historical bidding documents can be retrieved from the preset historical bidding project database. For example, taking civil engineering projects as an example, historical bidding documents corresponding to civil engineering projects can be retrieved from the preset historical bidding project database. The historical bidding documents record the historical comprehensive weights of each preset evaluation indicator corresponding to the civil engineering project. The historical comprehensive weights can be used as reference values ​​for this bid evaluation. The target comprehensive weights can be updated through interpolation calculation.

[0073] In one possible implementation, after extracting the project features from the project content in the bidding project documents, the method further includes: Obtain the preset weight adjustment rules for each feature item of the project. The preset weight adjustment rules record the comprehensive weight adjustment value for each feature item. Each feature item in the project features is judged according to the preset weight adjustment rules; When a feature item meets the preset weight adjustment rules, obtain the comprehensive weight adjustment value of the corresponding feature item; A correction coefficient vector is generated based on the comprehensive weight adjustment value, and the target comprehensive weight is adjusted based on the correction coefficient vector.

[0074] Here, project feature F reflects the special nature of the project content in the bidding project document. Therefore, based on the project feature, this embodiment of the invention designs a weight adjustment rule. The preset weight adjustment rule records the comprehensive weight adjustment value of each feature item, that is, the target comprehensive weight of the preset evaluation index is adaptively fine-tuned according to the project feature.

[0075] However, the trigger condition for adaptive fine-tuning is that the feature items meet the preset weight adjustment rules. Therefore, it is necessary to judge each feature item in the project features one by one according to the preset weight adjustment rules, and when the feature item meets the preset weight adjustment rules, obtain the comprehensive weight adjustment value of the corresponding feature item.

[0076] For example, the preset weight adjustment rules stipulate that: if the project type is civil engineering, the weight of the technical indicator needs to be increased by 3%, and the weight of the price indicator also needs to be increased by 3%; if the building area exceeds the preset area threshold, the weight of the credit indicator needs to be increased by 3%, and the weight of the price indicator needs to be decreased by 3%; if the number of floors exceeds the preset floor threshold, the weight of the credit indicator needs to be increased by 3%, and the weight of the price indicator needs to be increased by 3%; if the technical complexity level exceeds the preset level threshold, the weight of the technical indicator needs to be increased by 3%, and the weight of the price indicator needs to be decreased by 3%; if the construction period urgency is less than the preset number of days threshold, the weight of the price indicator increases by 3%. These preset weight adjustment rules can be adjusted according to the actual scenario, and the adjustment amount can be fine-tuned within a reasonable range. This is only an example and is not a limitation.

[0077] According to the preset weight adjustment rules, if a feature is found to meet the preset weight adjustment rules, such as the technical complexity level exceeding the preset level threshold, then the comprehensive weight adjustment value of the preset evaluation index in the preset weight adjustment rules can be obtained. This comprehensive weight adjustment value can be used as a correction coefficient to generate a correction coefficient vector. Used to adjust the overall weight of the target, denoted as 'm' represents the number of preset evaluation indicators. Furthermore, if multiple feature items satisfy the preset weight adjustment rules, multiple comprehensive weight adjustment values ​​are obtained accordingly. The comprehensive weight adjustment values ​​of the same preset evaluation indicator can be summed. Preset evaluation indicators that do not involve comprehensive weight adjustment values ​​have their correction coefficient vector... The vector value at the corresponding position is set to 0, and the resulting correction coefficient vector is... Notation: .

[0078] Finally, based on the correction coefficient vector Adjust the overall target weight The adjusted target comprehensive weight is .

[0079] In this embodiment of the invention, the adaptive adjustment of the target comprehensive weight is achieved through project characteristics. Reasonable weight adjustment rules are formulated according to the actual scenario of the project content. Theoretical evaluation and practical factors are combined to adjust the weight of each preset weight index, avoiding purely theoretical evaluation that is divorced from reality, so that the final evaluation result is more acceptable to the actual scenario.

[0080] When interpolating the target comprehensive weight based on the historical comprehensive weights in historical bidding documents to obtain the target dynamic weight, the target comprehensive weight used is the adjusted target comprehensive weight. .

[0081] In one possible implementation, the target comprehensive weight is calculated by interpolation based on the historical comprehensive weight of historical bidding documents to obtain the target dynamic weight. This can be achieved in the following ways, which are explained in detail below.

[0082] Obtain the similarity value when retrieving historical tender documents based on project features; The similarity values ​​are used as interpolation coefficients for interpolation calculations; The target dynamic weight is obtained by interpolating the historical comprehensive weight and the target comprehensive weight of the historical bidding documents based on the interpolation coefficient.

[0083] Here, when retrieving historical bidding project databases based on project feature F, the project features of historical bidding documents in the database are also extracted. Then, the similarity value (e.g., cosine similarity) between project feature F and this project feature is calculated, and finally, the historical bidding document with the highest cosine similarity is selected. Therefore, this similarity value is obtained and used as the interpolation coefficient for interpolation calculation, denoted as... According to the interpolation coefficients The historical comprehensive weight and target comprehensive weight of historical bidding documents can be interpolated to obtain the target dynamic weight, denoted as . Its specific calculation formula can be expressed as:

[0084] in, This refers to the historical comprehensive weight corresponding to the j-th preset evaluation indicator in the historical bidding documents.

[0085] In terms of the preset evaluation indicators, the target dynamic weights It can also be expressed as:

[0086] Where m represents the number of preset evaluation indicators, and k represents the k-th preset evaluation indicator. This represents the vector value corresponding to the k-th preset evaluation index in the correction coefficient vector. This represents the vector value corresponding to the j-th preset evaluation index in the correction coefficient vector.

[0087] It should be noted that the above embodiments describe adjusting the target comprehensive weight first using a correction coefficient vector, and then interpolating it using interpolation coefficients to obtain the target dynamic weight. However, in another implementation, the target comprehensive weight, adjusted based on the correction coefficient vector, can be directly used as the target dynamic weight without the need for interpolation calculations. In another implementation, the target comprehensive weights do not need to be adjusted through the correction coefficient vector and can be directly used for interpolation calculations. The results of the interpolation calculations can then be used as the target dynamic weights. .

[0088] This invention achieves fine-tuning of the target comprehensive weight by retrieving historical comprehensive weights from historical bidding documents. This utilizes historical bidding experience to assist in the review process, rather than relying on manual judgment, making the evaluation of weight allocation results more credible.

[0089] Step 104: Determine the risk weight of the bidder's bidding risk based on the bidder's historical bidding behavior data, and adjust the target dynamic weight based on the risk weight to obtain the final weight.

[0090] Here, this embodiment of the invention takes into account the differences in bidding behavior among bidders, which may lead to different bidding risks and significantly affect the final bid evaluation results. To address this, this embodiment of the invention also analyzes the historical bidding behavior data of the bidders and conducts a bidding risk assessment, incorporating the risk weights of bidding risks into the weighting evaluation system to eliminate the impact of these differences in bidding behavior on the bid evaluation results.

[0091] In one possible implementation, the risk weight of a bidder's bidding risk is determined based on the bidder's historical bidding behavior data. This can be achieved in the following ways, which are explained in detail below.

[0092] Bidding behavior identification is performed on the historical bidding behavior data of each bidder, resulting in multiple behavior types, including abnormal behavior of non-performance, performance behavior, and collusion behavior. For each type of behavior, a bidding risk assessment is conducted separately to obtain a risk level score; The risk level score is weighted based on a preset penalty coefficient, and the weighted results are summed to obtain the risk weight of the bidder's bidding risk.

[0093] Here, we first obtain historical bidding behavior data for each bidder. This data records various types of behavior generated by the bidder during the historical bidding process, including abnormal behavior such as non-performance, performance, and collusion. Performance refers to the bidder's normal behavior of strictly fulfilling the contract and carrying out the project after winning the bid in the historical evaluation. Abnormal behavior refers to the bidder's behavior such as failing to fulfill the contract, privately violating the contract, delaying or even refusing to carry out the project after winning the bid in the historical evaluation. Collusion refers to the bidder's behavior of colluding with other bidders in the historical evaluation. In addition, other behavior types can be added according to actual needs, which are not limited here.

[0094] Behavioral identification can determine the types of behaviors exhibited by bidders and count the frequency of each behavior type. These frequency counts are then used to assess bidding risk, mapping the occurrence counts to a risk level score. Generally, the frequency of normal behaviors such as contract performance is negatively correlated with the mapped risk level score, while the frequency of abnormal behaviors such as non-performance and bid-rigging is positively correlated with the mapped risk level score. For example, a higher frequency of abnormal behaviors results in a higher risk level score for that behavior type; a higher frequency of contract performance results in a lower risk level score; and a higher frequency of bid-rigging results in a higher risk level score.

[0095] For these behavioral types, this embodiment of the invention sets a penalty coefficient to weight these risk level scores, thus implementing "penalty," and sums the weighted results to obtain the personalized risk weight for each bidder's bidding risk. Since the evaluation dimensions of the bidding parties differ among the tendering parties, the penalty coefficient can be adjusted according to the actual situation. For example, some tendering parties cannot tolerate any collusion, so the corresponding penalty coefficient is set higher. The risk weight for a bidder's bidding risk is denoted as... The calculation formula is:

[0096] Where i represents the i-th bidder. , , These represent the risk levels of the i-th bidder in terms of non-performance, performance, and collusion. , as well as These represent the penalty coefficients set for non-compliance, compliance, and collusion, respectively. All three values ​​are positive, and the sum of the values ​​is 1.

[0097] Next, based on risk weights Dynamic weights for the target After making corrections, the final weights are obtained, denoted as... The calculation formula is expressed as:

[0098] in, This represents the final weight of the i-th bidder on the j-th preset evaluation indicator. This represents the preset dimension penalty sensitivity coefficient for the j-th preset evaluation index. The more sensitive the bidding party is to the bidding party's bidding behavior, the higher the sensitivity will be. The higher the value, the lower the value.

[0099] Step 105: Weight the initial evaluation score according to the final weight, and generate the bid evaluation result of the bidder on the bidding project documents based on the weighted comprehensive score.

[0100] The final weights determined through the weighting evaluation system outlined above incorporate subjective factors, objective factors, project characteristic factors (i.e., pre-set weighting adjustment rules), historical experience (i.e., historical bidding documents), and the bidders' bidding behavior, accurately measuring their overall performance during the bidding process. By weighting the initial evaluation scores based on these final weights, a comprehensive score measuring the bidders' overall performance can be obtained. However, since there are multiple bidders, the final weights for each bidder are determined first. Normalization is performed according to the preset evaluation indicators to obtain the personalized final weight, denoted as . , represented as:

[0101] in, This represents the personalized final weight of the i-th bidder on the j-th preset evaluation indicator, m represents the number of bidders, and k represents the k-th bidder.

[0102] The initial evaluation score is the decision matrix constructed in the above embodiments. Therefore, for the i-th bidder, its comprehensive score on a total of m preset evaluation indicators is... Represented as:

[0103] in, This represents the score corresponding to the j-th preset scoring indicator in the i-th bid document. In this embodiment of the invention, one bid document corresponds to one bidder. This represents the initial evaluation score corresponding to the j-th preset scoring indicator among the i-th bidders.

[0104] Overall score This can serve as a direct reference for the evaluation results, and the overall score will be used as a basis for the evaluation. The higher the score, the better the bidder's overall performance, and the more suitable they are for evaluation in this bidding process. Here, the overall score is used to determine the bidder's suitability. The bidders are sorted in reverse order based on their scores. This reverse-order sequence of bidders is the evaluation result of the bidding documents. In subsequent processes, the bidder with the highest score in the sequence can be directly selected as the bidding target, i.e., the bidder with the highest score is selected based on its overall score. The bidders are selected as the tendering targets. Furthermore, the bid documents can be supplemented with information detected during the initial evaluation score calculation process, including bid rigging, abnormal price values, non-standard terminology, and subsequent calculations of target dynamic weights, risk weights, and final weights. This generates a structured bid evaluation report for final manual confirmation by the tendering party to select the tendering targets.

[0105] In summary, this invention first evaluates the tender documents separately based on subjective and objective weights, and then synthesizes the results using least squares. This approach preserves the tenderer's prioritization of scoring indicators such as price and technology while accurately reflecting the objective differences in the tender documents. Furthermore, by extracting project characteristics and retrieving similar historical tender documents, it utilizes historical bidding experience to assist in the review process and adjusts the overall target weights without relying on manual judgment, making the weight allocation results more credible. In addition, by analyzing the bidders' historical bidding behavior to determine the risk weights for bidding, a "penalty" is imposed on bidders, thus correcting the weight allocation and eliminating the influence of diverse bids from multiple bidders. This ensures that the final weights accurately measure the bidders' overall performance, making the evaluation results more standardized and consistent. Moreover, the entire evaluation process can be automated, without any manual operation or reliance on human judgment, effectively improving the reliability of the evaluation results.

[0106] The automated bid evaluation device for bidding projects provided by this invention will be described in detail below.

[0107] like Figure 2 As shown, the automated bid evaluation device for bidding projects specifically includes: an acquisition module 201, a determination module 202, a calculation module 203, a correction module 204, and an evaluation module 205.

[0108] Specifically, the acquisition module 201 is used to acquire the initial evaluation scores of the bidding project documents of the bidding party and the bid documents of the bidding party after review; the determination module 202 is used to determine the subjective weight of the bidding party on the corresponding scoring indicators of the bidding project documents, determine the objective weight of the bidding party on the preset evaluation indicators, and perform least squares calculation on the subjective weight and the objective weight to obtain the target comprehensive weight; the calculation module 203 is used to extract the project features of the project content in the bidding project documents, retrieve historical bidding documents from the preset historical bidding project database based on the project features, and interpolate the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight; the correction module 204 is used to determine the risk weight of the bidding party's bidding risk based on the bidding party's historical bidding behavior data, and correct the target dynamic weight based on the risk weight to obtain the final weight; the evaluation module 205 is used to weight the initial evaluation score according to the final weight, and generate the evaluation result of the bidding party on the bidding project documents based on the weighted comprehensive score.

[0109] In one possible implementation, the calculation module is further configured to: after extracting the project features from the project content in the bidding project file, obtain a preset weight adjustment rule for each feature item of the project features, wherein the preset weight adjustment rule records a comprehensive weight adjustment value for each feature item; determine each feature item in the project features according to the preset weight adjustment rule; when a feature item satisfies the preset weight adjustment rule, obtain the comprehensive weight adjustment value of the corresponding feature item; generate a correction coefficient vector based on the comprehensive weight adjustment value, and adjust the target comprehensive weight based on the correction coefficient vector.

[0110] The automated bid evaluation device for bidding projects provided in the above embodiments can realize the technical solutions described in the embodiments of the automated bid evaluation method for bidding projects. The specific implementation principles of each module or unit can be found in the corresponding content in the embodiments of the automated bid evaluation method for bidding projects, and their technical effects can also be referred to each other. They will not be repeated here.

[0111] like Figure 3 As shown, the present invention also provides an electronic device 300. The electronic device 300 includes a processor 301, a memory 302, and a display 303. Figure 3 Only some components of the electronic device 300 are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0112] In some embodiments, memory 302 may be an internal storage unit of electronic device 300, such as a hard disk or memory of electronic device 300. In other embodiments, memory 302 may also be an external storage device of electronic device 300, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on electronic device 300.

[0113] Furthermore, the memory 302 may include both internal storage units of the electronic device 300 and external storage devices. The memory 302 is used to store application software and various types of data installed on the electronic device 300.

[0114] In some embodiments, processor 301 may be a central processing unit (CPU), microprocessor, or other data processing chip, used to run program code stored in memory 302 or process data, such as the automated bid evaluation method for bidding projects in this invention.

[0115] In some embodiments, display 303 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 303 is used to display information from electronic device 300 and to display a visual user interface. Components 301-303 of electronic device 300 communicate with each other via a system bus.

[0116] In some embodiments of the present invention, when the processor 301 executes the computer program in the memory 302, the following steps can be implemented: obtaining the bidding project documents of the bidding party and the initial evaluation scores of the bid documents of the bidding party after bid review; determining the subjective weight of the bidding party on the preset scoring indicators corresponding to the bidding project documents, determining the objective weight of the bidding party on the preset evaluation indicators, and performing least squares calculation on the subjective weight and the objective weight to obtain the target comprehensive weight; extracting the project features of the project content in the bidding project documents, retrieving historical bidding documents from the preset historical bidding project database based on the project features, and interpolating the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight; determining the risk weight of the bidding party's bidding risk based on the bidding party's historical bidding behavior data, and correcting the target dynamic weight based on the risk weight to obtain the final weight; weighting the initial evaluation score according to the final weight, and generating the bid evaluation result of the bidding party on the bidding project documents based on the weighted comprehensive score.

[0117] It should be understood that when the processor 301 executes the computer program in the memory 302, in addition to the functions described above, it can also perform other functions, as can be found in the description of the corresponding method embodiments above.

[0118] Furthermore, the embodiments of the present invention do not specifically limit the type of electronic device 300 mentioned. Electronic device 300 can be a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, laptop computer, or other portable electronic device. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic device can also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, electronic device 300 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0119] In another aspect, the present invention also provides an automated bid evaluation system for bidding projects, including a data interface device, a communication device, and the electronic device 300 in the above embodiments. The data interface device communicates with the electronic device 300 through the communication device.

[0120] The data interface device is used to receive and obtain the bidding project documents from the bidding party and the bid documents from the bidders; the communication device is used to transmit the bidding project documents and bid documents to the electronic device 300, and after the processor 301 of the electronic device 300 executes the program stored in the memory 302, it transmits the evaluation results of the output bidding project documents to the data interface device.

[0121] In some embodiments, the data interface device may include general peripheral connection interfaces (such as USB interfaces), audio and video transmission interfaces (such as high-definition multimedia interfaces, analog or digital interfaces), storage device interfaces (such as serial SATA interfaces, solid-state drive interfaces), industrial and serial communication interfaces (such as RS-232, RS-485), network communication interfaces (such as fiber optic interfaces, router converters), etc.

[0122] In some embodiments, the communication equipment may be network communication equipment deployed with network links such as the Internet or Ethernet, such as switches, routers, firewalls, gateways, bridges, repeaters, wireless access points (APs), modems, optical transceivers, and fiber optic transceivers.

[0123] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0124] The above provides a detailed description of the automated bid evaluation method, apparatus, equipment, and system for bidding projects provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. An automated bid evaluation method for bidding projects, characterized in that, include: Obtain the bidding project documents from the bidding party and the initial evaluation scores of the bid documents after bid review from the bidders; Determine the subjective weight of the tendering party on the preset scoring indicators corresponding to the tender project documents, determine the objective weight of the bidder on the preset evaluation indicators, and perform least squares calculation on the subjective weight and the objective weight to obtain the target comprehensive weight. Extract project features from the project content in the bidding project documents, retrieve historical bidding documents from a preset historical bidding project database based on the project features, and interpolate the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight; The risk weight for the bidder's bidding risk is determined based on the bidder's historical bidding behavior data, and the target dynamic weight is adjusted based on the risk weight to obtain the final weight. The initial evaluation score is weighted according to the final weight, and the bid evaluation result of the bidder on the bidding project documents is generated based on the weighted comprehensive score.

2. The automated bid evaluation method for bidding projects according to claim 1, characterized in that, The bidding documents were obtained through the following methods: Obtain the drawings and documents for the bidding project; The drawing file is parsed for content elements, and the parsed file is then checked for content integrity. The quantities of the project content in the file after integrity verification are calculated, and the isolated forest model is called to detect anomalies in the calculated quantities. When the anomaly detection result is no anomaly, a cost database for the project content is constructed; The cost database, the project quantity, and the integrity-verified file are bound together to obtain the bidding project document.

3. The automated bid evaluation method for bidding projects according to claim 1, characterized in that, The preset scoring indicators include: credit indicators, technical indicators, content indicators, price indicators, and balance indicators. The initial evaluation score is obtained by calculating the score of the tender document on the preset scoring indicators as the initial evaluation score. The calculation of the scores of the bid documents on the preset scoring indicators includes: The tender documents are subject to qualification review to obtain a qualification score on the creditworthiness indicator. The tender document is segmented into words, and the segmented content is fuzzy matched based on a preset terminology database to obtain a technical standardization score in terms of technical indicators. The semantic similarity of the content of the tender documents is calculated to obtain a bid-rigging risk score on the content indicator. Calculate the matching degree between the price value and the corresponding budget value of each item in the tender document to obtain the price deviation score on the price indicator; The TOPSIS multi-criteria decision model is used to calculate the relative closeness between the bid prices recorded in each tender document and the benchmark price, and the bid imbalance score on the balance index is obtained.

4. The automated bid evaluation method for bidding projects according to claim 1, characterized in that, The step of performing least squares calculation on the subjective weights and the objective weights to obtain the target comprehensive weights includes: Construct a least squares model, wherein the objective function of the least squares model is the sum of squares of a first difference and a second difference, the first difference being the difference between the target comprehensive weight and the subjective weight, and the second difference being the difference between the target comprehensive weight and the objective weight; The least squares model is optimized with minimizing the sum of squares as the optimization objective to obtain the target comprehensive weight.

5. The automated bid evaluation method for bidding projects according to claim 1, characterized in that, After extracting the project features from the project content in the bidding project document, the method further includes: Obtain a preset weight adjustment rule for each feature item of the project features, wherein the preset weight adjustment rule records the comprehensive weight adjustment value for each feature item; Each feature item in the project features is judged according to the preset weight adjustment rule; When a feature item meets the preset weight adjustment rules, obtain the comprehensive weight adjustment value of the corresponding feature item; A correction coefficient vector is generated based on the comprehensive weight adjustment value, and the target comprehensive weight is adjusted based on the correction coefficient vector.

6. The automated bid evaluation method for bidding projects according to claim 1, characterized in that, The step of interpolating the target comprehensive weight based on the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight includes: Obtain the similarity value when retrieving the historical tender documents based on the project features; The similarity value is determined as the interpolation coefficient for interpolation calculation; The target dynamic weight is obtained by interpolating the historical comprehensive weight of the historical bidding documents and the target comprehensive weight based on the interpolation coefficient.

7. The automated bid evaluation method for bidding projects according to claim 1, characterized in that, The risk weight for determining bidding risks based on the bidder's historical bidding behavior data includes: Bidding behavior identification is performed on the historical bidding behavior data of each bidder, resulting in multiple behavior types, including abnormal behavior of non-performance, performance behavior, and collusion behavior. For each type of behavior, a bidding risk assessment is conducted separately to obtain a risk level score; The risk level score is weighted based on a preset penalty coefficient, and the weighted results are summed to obtain the risk weight of the bidder's bidding risk.

8. An automated bid evaluation device for bidding projects, characterized in that, include: The acquisition module is used to acquire the bidding project documents from the bidding party and the initial evaluation scores of the bidders' bid documents after review. The determination module is used to determine the subjective weight of the tendering party on the corresponding scoring indicators in the tender project document, determine the objective weight of the bidder on the preset evaluation indicators, and perform least squares calculation on the subjective weight and the objective weight to obtain the target comprehensive weight. The calculation module is used to extract project features from the project content in the bidding project file, retrieve historical bidding documents from a preset historical bidding project database based on the project features, and interpolate the target comprehensive weight according to the historical comprehensive weight of the historical bidding documents to obtain the target dynamic weight. The correction module is used to determine the risk weight of the bidder's bidding risk based on the bidder's historical bidding behavior data, and to correct the target dynamic weight based on the risk weight to obtain the final weight; The bid evaluation module is used to weight the initial evaluation score according to the final weight, and generate the bid evaluation result of the bidder on the bidding project documents based on the weighted comprehensive score.

9. An electronic device, characterized in that, Including memory and processor, among which, The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps of the automated bid evaluation method for bidding projects as described in any one of claims 1 to 7.

10. An automated bid evaluation system for bidding projects, characterized in that, It includes a data interface device, a communication device, and an electronic device as described in claim 9, wherein the data interface device communicates with the electronic device through the communication device; The data interface device is used to receive and obtain the bidding project documents from the bidding party and the bid documents from the bidders. The communication device is used to transmit the bidding project documents and the bid documents to the electronic device, and after the processor of the electronic device executes the program stored in the memory, it transmits the evaluation results of the output bidding project documents to the data interface device.