Intelligent decision support system for whole bidding and tendering process

Abstract

The invention relates to the technical field of bidding and tendering decisions, and discloses an intelligent decision support system for a whole bidding and tendering process, a model construction module analyzes the whole bidding and tendering process to obtain nodes of the whole bidding and tendering process, constructs a node model of the whole bidding and tendering process, and outputs corresponding node evaluation values; a first calculation module determines a node evaluation value set according to all node evaluation values, and calculates corresponding node evaluation fluctuation metric values; a second calculation module analyzes the evaluation fluctuation metric values of all the nodes and calculates a bidding and tendering difference coefficient; the process decision-making module performs intelligent decision-making on the whole bidding and tendering process according to the bidding and tendering difference coefficient and a preset bidding and tendering difference coefficient, calculates a quantitative index capable of representing the whole health degree of the whole process through deep mining and analysis of each node evaluation value, provides clear and reliable decision-making support for management personnel, and improves the management efficiency. The intelligent and precise bidding management is realized, and the project success rate and the resource utilization efficiency are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of bidding and tendering decision-making technology, and more specifically, to an intelligent decision support system for the entire bidding and tendering process. Background Technology

[0002] Tendering and bidding is the primary method of project contracting and subcontracting in a market economy, widely used in engineering construction, goods procurement, and service procurement. A complete tendering and bidding process typically includes several key stages such as tender planning, pre-qualification, tender document preparation, bidding, bid opening, bid evaluation, bid awarding, and contract signing. This process demands a high degree of complexity, standardization, and professionalism; any oversight or decision-making error at any stage can lead to serious consequences such as failed tendering, increased project costs, project delays, legal disputes, or even the loss of state-owned assets.

[0003] In traditional bidding and tendering management practices, decision support relies heavily on the personal experience and subjective judgment of project managers. For example, when evaluating the quality of tender documents, they are usually reviewed manually by experts. However, the evaluation standards are difficult to standardize and are easily influenced by personal preferences and energy levels, resulting in inconsistent evaluation results and low efficiency. Furthermore, for monitoring the entire process, traditional methods are often limited to isolated, qualitative checks of each node, lacking a systematic and quantitative approach to dynamically perceive and evaluate the operational status and health of the entire process. Summary of the Invention

[0004] This invention provides an intelligent decision support system for the entire bidding process. It constructs a precise node evaluation model and calculates quantitative indicators that characterize the overall health of the entire process through in-depth mining and analysis of the evaluation values ​​of each node. Based on these indicators, it provides managers with clear and reliable decision support, thereby achieving intelligent and precise bidding management and effectively improving project success rate and resource utilization efficiency.

[0005] To achieve the above objectives, the present invention provides an intelligent decision support system for the entire bidding process, comprising:

[0006] The model building module is used to analyze the entire bidding process, obtain the nodes of the entire bidding process, build a bidding process node model for each bidding process node, and output the node evaluation value of the corresponding bidding process node based on the bidding process node model.

[0007] The first calculation module is used to determine the set of node evaluation values ​​based on all node evaluation values, and to calculate the node evaluation fluctuation metric value corresponding to each node evaluation value based on the set of node evaluation values.

[0008] The second calculation module is used to analyze the fluctuation metric values ​​of all nodes and calculate the bidding difference coefficient of the entire bidding process based on the analysis results.

[0009] The process decision module is used to pre-set a preset bidding difference coefficient and make intelligent decisions on the entire bidding process based on the relationship between the bidding difference coefficient and the preset bidding difference coefficient.

[0010] Furthermore, the model building module is used for:

[0011] Obtain a historical tender document dataset, which includes the original text data of the tender documents and their corresponding key project indicators;

[0012] Demand features are extracted from the original text data, including explicitness features, completeness features, and compliance features;

[0013] Based on the key indicators of the project, a comprehensive score label is generated for the historical bidding document dataset;

[0014] Using the aforementioned demand features as input and the aforementioned comprehensive rating labels as training targets, a machine learning model is trained.

[0015] Verify and output the trained full-process node model of bidding.

[0016] Furthermore, the model building module is used for:

[0017] The required features include explicitness feature extraction, completeness feature extraction, and compliance feature extraction;

[0018] The explicit feature extraction includes using a natural language processing model to calculate the perplexity of the requirement description text, and / or identifying and counting the frequency of ambiguous words in the text;

[0019] The integrity feature extraction includes calculating the coverage of the original text data to the feature list based on a predefined feature list, using text matching or entity recognition technology.

[0020] The compliance feature extraction includes comparing the semantic similarity of the original text data with related texts to identify clauses with compliance risks.

[0021] Furthermore, the first computing module is used for:

[0022] Determine the range of node evaluation values ​​corresponding to the set of node evaluation values, wherein the range of node evaluation values ​​includes a first preset node evaluation value and a second preset node evaluation value, and the first preset node evaluation value is less than the second preset node evaluation value;

[0023] The set of node evaluation values ​​is traversed. When a node evaluation value is less than the first preset node evaluation value, the difference between the first preset node evaluation value and the node evaluation value is calculated and used as a node evaluation fluctuation metric.

[0024] When the node evaluation value is greater than the second preset node evaluation value, the difference between the node evaluation value and the second preset node evaluation value is calculated and used as the node evaluation fluctuation metric.

[0025] When the node evaluation value is within the range of the node evaluation value, the first difference between the node evaluation value and the first preset node evaluation value is calculated, the second difference between the second preset node evaluation value and the node evaluation value is calculated, and the average of the first difference and the second difference is used as the node evaluation fluctuation metric.

[0026] Furthermore, the second computing module is used for:

[0027] The fluctuation metric values ​​of every two nodes are combined to obtain multiple sets of node fluctuation metric values, and the relative fluctuation difference is calculated based on the set of node fluctuation metric values.

[0028] The bidding difference coefficient for the entire bidding process is calculated based on the relative fluctuation difference.

[0029] Furthermore, the second computing module is used for:

[0030] Calculate the first difference of the node evaluation volatility metric group, wherein the first difference is the absolute value of the difference between the two node evaluation volatility metrics in the node evaluation volatility metric group.

[0031] Extract the maximum node evaluation volatility metric and the minimum node evaluation volatility metric, and calculate the second difference between the maximum node evaluation volatility metric and the minimum node evaluation volatility metric;

[0032] Calculate the ratio of the second difference to each of the first differences, and use it as the relative fluctuation difference.

[0033] Furthermore, the second computing module is used for:

[0034] Extract the same relative fluctuation difference from the relative fluctuation difference, and calculate the stability difference coefficient of the entire bidding process based on the same relative fluctuation difference;

[0035] Extract the remaining relative fluctuation difference from the relative fluctuation difference, and calculate the fluctuation difference coefficient of the entire bidding process based on the remaining relative fluctuation difference;

[0036] The weighted sum of the stability difference coefficient and the wave state difference coefficient yields the bidding difference coefficient for the entire bidding process.

[0037] Furthermore, the second computing module is used for:

[0038] Based on the same relative volatility difference, multiple relative volatility difference sequences are obtained;

[0039] The number of first relative volatility difference sequences in the statistical relative volatility difference sequence;

[0040] Extract one relative volatility difference from each of the relative volatility difference sequences and calculate the sum of the first relative volatility differences;

[0041] Calculate the mean of relative volatility differences for all identical relative volatility differences, remove all relative volatility difference sequences that are less than the mean of relative volatility differences, and count the number of second relative volatility difference sequences among the remaining relative volatility difference sequences.

[0042] Extract one relative volatility difference from each of the remaining relative volatility difference sequences, and calculate the second relative volatility difference sum value;

[0043] The stability difference coefficient for the entire bidding process is calculated based on the number of the first relative fluctuation difference sequence, the number of the second relative fluctuation difference sequence, the sum of the first relative fluctuation difference, and the sum of the second relative fluctuation difference.

[0044]

[0045] Where s is the stability difference coefficient of the entire bidding process, r1 is the number of the first relative fluctuation difference sequence, r2 is the number of the second relative fluctuation difference sequence, t1 is the sum of the first relative fluctuation difference, and t2 is the sum of the second relative fluctuation difference.

[0046] Furthermore, the second computing module is used for:

[0047] The waveform difference coefficient for the entire bidding process is calculated using the following formula:

[0048]

[0049] Where w is the volatility difference coefficient of the entire bidding process, n is the number of remaining relative volatility differences, and g i For the relative fluctuation difference of the i-th remaining value, g i+1 This represents the relative fluctuation difference of the (i+1)th remaining variable.

[0050] Furthermore, the process decision module is used for:

[0051] When the bidding difference coefficient is less than the preset bidding difference coefficient, it is determined that the entire bidding process meets the bidding conditions.

[0052] When the bidding difference coefficient is greater than or equal to the preset bidding difference coefficient, it is determined that the entire bidding process does not meet the bidding conditions.

[0053] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0054] This invention discloses an intelligent decision support system for the entire bidding process. The model building module analyzes the entire bidding process, obtains the nodes, constructs a node model, and outputs corresponding node evaluation values. The first calculation module determines the set of node evaluation values ​​based on all node evaluation values ​​and calculates the corresponding node evaluation fluctuation metric. The second calculation module analyzes all node evaluation fluctuation metrics and calculates the bidding difference coefficient. The process decision module makes intelligent decisions for the entire bidding process based on the bidding difference coefficient and a preset bidding difference coefficient. Through in-depth mining and analysis of the evaluation values ​​of each node, it calculates quantitative indicators that characterize the overall health of the entire process, providing managers with clear and reliable decision support, achieving intelligent and precise bidding management, and effectively improving project success rate and resource utilization efficiency. Attached Figure Description

[0055] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0056] Figure 1 The diagram shows a schematic representation of an intelligent decision support system for the entire bidding process according to an embodiment of the present invention. Detailed Implementation

[0057] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0058] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0059] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0060] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0061] The following is a description of preferred embodiments of the present invention in conjunction with the accompanying drawings.

[0062] like Figure 1 As shown, an embodiment of the present invention discloses an intelligent decision support system for the entire bidding process, comprising: a model building module, a first calculation module, a second calculation module, and a process decision module.

[0063] In some embodiments of this application, the model building module is used to analyze the entire bidding process, obtain the nodes of the entire bidding process, build a bidding process node model for each bidding process node, and output the node evaluation value of the corresponding bidding process node according to the bidding process node model.

[0064] In this embodiment, the entire bidding process includes bidding requirements analysis, prequalification, bid document review, clarification and negotiation, winning bid decision and contract performance tracking.

[0065] In some embodiments of this application, the model building module is used for:

[0066] Obtain a historical tender document dataset, which includes the original text data of the tender documents and their corresponding key project indicators;

[0067] Demand features are extracted from the original text data, including explicitness features, completeness features, and compliance features;

[0068] Based on the key indicators of the project, a comprehensive score label is generated for the historical bidding document dataset;

[0069] Using the aforementioned demand features as input and the aforementioned comprehensive rating labels as training targets, a machine learning model is trained.

[0070] Verify and output the trained full-process node model of bidding.

[0071] In this embodiment, the historical bidding document dataset clearly defines its source and nature, representing a collection of documents from real bidding projects that have already occurred. Raw text data refers to the unprocessed or minimally processed content of documents such as the full text of bidding announcements, requirements specifications, technical specifications, and contract terms. This retains the richest information. Key project metrics are real-world results data that provide the model with the "target" for learning. These metrics are used to measure the final execution effect of the bidding project, such as: the number of valid bidders (reflecting competitiveness), the deviation rate between the winning bid and the budget (reflecting cost control), the number of changes during contract performance (reflecting the clarity of requirements), and the number of failed bids.

[0072] In this embodiment, the core operation of extracting requirement features aims to quantify the quality of the tender documents. **Clarity Features:** This measures whether the requirement description is clear and unambiguous. For example, NLP techniques are used to count the number of vague terms (such as "advanced" or "reasonable"), or to analyze sentence complexity and confusion. **Completeness Features:** This measures whether the requirements cover all necessary aspects. For example, coverage is calculated by checking whether key sections such as technical parameters, acceptance criteria, payment methods, and after-sales service are included. **Compliance Features:** This measures whether the tender documents comply with laws, regulations, and industry standards. For example, text matching or semantic similarity calculations are used to identify whether exclusive or discriminatory clauses exist in the documents.

[0073] In this embodiment, key project indicators (such as competitiveness and cost control effectiveness) are used to infer the quality of the tender document itself. A high-quality tender document (with clear, complete, and compliant requirements) typically leads to good project outcomes (such as sufficient competition and controllable costs). Therefore, these outcome indicators can be used to calculate a "comprehensive score" (i.e., a label) for each historical tender document. For example, multiple key indicators can be normalized and then weighted to obtain a total score. This score is the target that the model needs to predict.

[0074] In this embodiment, verification is performed by using a dataset that was not used in training (validation set or test set) to evaluate the performance of the trained model, ensuring its accuracy and generalization ability.

[0075] In this embodiment, based on the above model training method, the bidding process node model corresponding to each bidding process node can be obtained, and then the node evaluation value corresponding to each bidding process node can be output.

[0076] The beneficial effects of the above technical solution are as follows: by extracting demand features from historical bidding documents and generating comprehensive scoring labels using key project indicators, and then training a machine learning model, accurate evaluation of all nodes in the bidding process can be achieved. This data-driven model building approach not only improves the objectivity and accuracy of the evaluation but also enables the model to automatically learn and adapt to the characteristics of different types of bidding projects, thereby enhancing the system's versatility and flexibility. The verification process further ensures the reliability and stability of the model, providing a solid foundation for subsequent intelligent decision-making.

[0077] In some embodiments of this application, the model building module is used for:

[0078] The required features include explicitness feature extraction, completeness feature extraction, and compliance feature extraction;

[0079] The explicit feature extraction includes using a natural language processing model to calculate the perplexity of the requirement description text, and / or identifying and counting the frequency of ambiguous words in the text;

[0080] The integrity feature extraction includes calculating the coverage of the original text data to the feature list based on a predefined feature list, using text matching or entity recognition technology.

[0081] The compliance feature extraction includes comparing the semantic similarity of the original text data with related texts to identify clauses with compliance risks.

[0082] The beneficial effects of the above technical solution are as follows: By refining the extraction process of requirement features, the clarity feature extraction utilizes a natural language processing model to calculate perplexity and the frequency of ambiguous words, enabling a more accurate quantification of the clarity of the requirement description; the completeness feature extraction is based on a predefined list of elements for text matching or entity recognition, ensuring a comprehensive assessment of requirement coverage; and the compliance feature extraction identifies compliance risk clauses through semantic similarity comparison, effectively improving the efficiency of compliance review of bidding documents. This enhances the accuracy and reliability of the model's evaluation of all nodes in the bidding process, providing more solid data support for subsequent intelligent decision-making.

[0083] In some embodiments of this application, the first calculation module is used to determine a set of node evaluation values ​​based on all node evaluation values, and to calculate a node evaluation fluctuation metric value corresponding to each node evaluation value based on the set of node evaluation values.

[0084] In some embodiments of this application, the first computing module is used for:

[0085] Determine the range of node evaluation values ​​corresponding to the set of node evaluation values, wherein the range of node evaluation values ​​includes a first preset node evaluation value and a second preset node evaluation value, and the first preset node evaluation value is less than the second preset node evaluation value;

[0086] The set of node evaluation values ​​is traversed. When a node evaluation value is less than the first preset node evaluation value, the difference between the first preset node evaluation value and the node evaluation value is calculated and used as a node evaluation fluctuation metric.

[0087] When the node evaluation value is greater than the second preset node evaluation value, the difference between the node evaluation value and the second preset node evaluation value is calculated and used as the node evaluation fluctuation metric.

[0088] When the node evaluation value is within the range of the node evaluation value, the first difference between the node evaluation value and the first preset node evaluation value is calculated, the second difference between the second preset node evaluation value and the node evaluation value is calculated, and the average of the first difference and the second difference is used as the node evaluation fluctuation metric.

[0089] In this embodiment, the node evaluation value range is a pre-set range within which the node evaluation values ​​are relatively stable. The node evaluation value range is preferably [5,8], but can be adjusted adaptively according to actual needs.

[0090] The beneficial effects of the above technical solution are: by setting a range for node evaluation values ​​and calculating corresponding node evaluation fluctuation metrics for node evaluation values ​​under different conditions, the deviation of each node evaluation value from the preset range can be characterized more precisely. It considers not only cases where node evaluation values ​​exceed the range but also fluctuations within the range, enabling the node evaluation fluctuation metrics to more comprehensively and accurately reflect the stability and volatility of node evaluation values.

[0091] In some embodiments of this application, the second calculation module is used to analyze the fluctuation metric values ​​of all nodes and calculate the bidding difference coefficient of the entire bidding process based on the analysis results.

[0092] In some embodiments of this application, the second computing module is used for:

[0093] The fluctuation metric values ​​of every two nodes are combined to obtain multiple sets of node fluctuation metric values, and the relative fluctuation difference is calculated based on the set of node fluctuation metric values.

[0094] The bidding difference coefficient for the entire bidding process is calculated based on the relative fluctuation difference.

[0095] In some embodiments of this application, the second computing module is used for:

[0096] Calculate the first difference of the node evaluation volatility metric group, wherein the first difference is the absolute value of the difference between the two node evaluation volatility metrics in the node evaluation volatility metric group.

[0097] Extract the maximum node evaluation volatility metric and the minimum node evaluation volatility metric, and calculate the second difference between the maximum node evaluation volatility metric and the minimum node evaluation volatility metric;

[0098] Calculate the ratio of the second difference to each of the first differences, and use it as the relative fluctuation difference.

[0099] In this embodiment, if there are individual nodes that are not combined and their fluctuation metrics are evaluated, they can be deleted.

[0100] The beneficial effects of the above technical solution are as follows: by combining the fluctuation metrics of each node pairwise and calculating the relative fluctuation difference, the fluctuation relationships between different nodes can be captured more meticulously. The calculation of the absolute value of the first difference ensures the independence of the fluctuation direction, while the ratio processing of the second difference to the first difference further amplifies the impact of extreme fluctuations on the overall process. This not only preserves the fluctuation characteristics of the original data but also enhances the comparability between projects of different sizes through relativization. The final generated bidding difference coefficient, as a quantitative indicator of the overall process health, can intuitively reflect the degree of collaborative stability between nodes, providing a scientific basis for managers to identify potential risk points and optimize process configuration.

[0101] In some embodiments of this application, the second computing module is used for:

[0102] Extract the same relative fluctuation difference from the relative fluctuation difference, and calculate the stability difference coefficient of the entire bidding process based on the same relative fluctuation difference;

[0103] Extract the remaining relative fluctuation difference from the relative fluctuation difference, and calculate the fluctuation difference coefficient of the entire bidding process based on the remaining relative fluctuation difference;

[0104] The weighted sum of the stability difference coefficient and the wave state difference coefficient yields the bidding difference coefficient for the entire bidding process.

[0105] In this embodiment, the stability difference coefficient and the wave state difference coefficient are weighted based on the subjective weighting method and the objective weighting method. The weight of the stability difference coefficient is preferably 0.6, and the weight of the wave state difference coefficient is preferably 0.4.

[0106] The beneficial effects of the above technical solution are as follows: by subdividing relative fluctuation differences into two dimensions—stable differences and volatile differences—and calculating the corresponding stable difference coefficient and volatile difference coefficient respectively, a hierarchical analysis of the fluctuation characteristics of the entire bidding process is achieved. The stable difference coefficient focuses on repetitive fluctuation patterns and can identify inherent stability problems in the process; the volatile difference coefficient captures non-repetitive fluctuation characteristics and effectively reflects the dynamic change risks in the process. A weighted summation method combining subjective and objective weighting is adopted, considering both the actual importance of different difference coefficients to the entire process and the statistical characteristics of the data itself. The bidding difference coefficient reflects the overall fluctuation status of the entire bidding process, becoming a core indicator for measuring the collaborative effectiveness of the entire process.

[0107] In some embodiments of this application, the second computing module is used for:

[0108] Based on the same relative volatility difference, multiple relative volatility difference sequences are obtained;

[0109] The number of first relative volatility difference sequences in the statistical relative volatility difference sequence;

[0110] Extract one relative volatility difference from each of the relative volatility difference sequences and calculate the sum of the first relative volatility differences;

[0111] Calculate the mean of relative volatility differences for all identical relative volatility differences, remove all relative volatility difference sequences that are less than the mean of relative volatility differences, and count the number of second relative volatility difference sequences among the remaining relative volatility difference sequences.

[0112] Extract one relative volatility difference from each of the remaining relative volatility difference sequences, and calculate the second relative volatility difference sum value;

[0113] The stability difference coefficient for the entire bidding process is calculated based on the number of the first relative fluctuation difference sequence, the number of the second relative fluctuation difference sequence, the sum of the first relative fluctuation difference, and the sum of the second relative fluctuation difference.

[0114]

[0115] Where s is the stability difference coefficient of the entire bidding process, r1 is the number of the first relative fluctuation difference sequence, r2 is the number of the second relative fluctuation difference sequence, t1 is the sum of the first relative fluctuation difference, and t2 is the sum of the second relative fluctuation difference.

[0116] In this embodiment, the number of relative fluctuation differences in each relative fluctuation difference sequence is greater than or equal to 2, and the relative fluctuation differences between each relative fluctuation difference sequence are different.

[0117] The beneficial effects of the above technical solution are: by constructing a relative fluctuation difference sequence and implementing a multi-level screening mechanism, it effectively distinguishes between stable fluctuation patterns and random fluctuation interference in the bidding process. It achieves a quantitative assessment of repetitive fluctuation characteristics in the process, providing a reliable basis for identifying process inertia risks and optimizing node coordination mechanisms.

[0118] In some embodiments of this application, the second computing module is used for:

[0119] The waveform difference coefficient for the entire bidding process is calculated using the following formula:

[0120]

[0121] Where w is the volatility difference coefficient of the entire bidding process, n is the number of remaining relative volatility differences, and g i For the relative fluctuation difference of the i-th remaining value, g i+1 This represents the relative fluctuation difference of the (i+1)th remaining variable.

[0122] The beneficial effects of the above technical solution are as follows: by employing the method of calculating the cumulative difference of adjacent relative fluctuations, the wave pattern difference coefficient can accurately capture the non-repetitive fluctuation characteristics throughout the entire bidding process. It not only reflects the time-series trend of fluctuations but also eliminates directional interference through the processing of the absolute values ​​of adjacent differences, making the wave pattern difference coefficient an effective indicator for measuring the risk of dynamic changes in the process.

[0123] In some embodiments of this application, the process decision module is used to pre-set a preset bidding difference coefficient and make intelligent decisions on the entire bidding process based on the relationship between the bidding difference coefficient and the preset bidding difference coefficient.

[0124] In some embodiments of this application, the process decision module is used for:

[0125] When the bidding difference coefficient is less than the preset bidding difference coefficient, it is determined that the entire bidding process meets the bidding conditions.

[0126] When the bidding difference coefficient is greater than or equal to the preset bidding difference coefficient, it is determined that the entire bidding process does not meet the bidding conditions.

[0127] In this embodiment, the preset bidding difference coefficient is preferably 5, but it can be adjusted adaptively according to actual needs.

[0128] The beneficial effects of the above technical solution are as follows: by setting a preset bidding difference coefficient as a threshold standard for process health, the system achieves automated compliance judgment of the entire bidding process. When the actual calculated bidding difference coefficient is lower than this threshold, the system automatically determines that the process is under control; otherwise, it triggers an early warning mechanism. This decision-making method based on quantitative indicators not only overcomes the subjective bias of traditional manual review but also enhances the system's adaptability through a dynamic threshold adjustment mechanism (such as optimizing preset values ​​based on parameters like project type and scale). The resulting intelligent decision-making results can provide bidding parties with a scientific basis for risk assessment and simultaneously build a data-driven process monitoring system for regulatory authorities.

[0129] In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0130] Although the invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the embodiments disclosed in this invention can be combined with each other in any way. The fact that not all of these combinations are described in this specification is merely for the sake of brevity and resource conservation.

[0131] It will be understood by those skilled in the art that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An intelligent decision support system for the entire bidding process, characterized in that, include: The model building module is used to analyze the entire bidding process, obtain the nodes of the entire bidding process, build a bidding process node model for each bidding process node, and output the node evaluation value of the corresponding bidding process node based on the bidding process node model. The first calculation module is used to determine the set of node evaluation values ​​based on all node evaluation values, and to calculate the node evaluation fluctuation metric value corresponding to each node evaluation value based on the set of node evaluation values. The second calculation module is used to analyze the fluctuation metric values ​​of all nodes and calculate the bidding difference coefficient of the entire bidding process based on the analysis results. The process decision module is used to pre-set a preset bidding difference coefficient and make intelligent decisions on the entire bidding process based on the relationship between the bidding difference coefficient and the preset bidding difference coefficient.

2. The intelligent decision support system for the entire bidding process according to claim 1, characterized in that, The model building module is used for: Obtain a historical tender document dataset, which includes the original text data of the tender documents and their corresponding key project indicators; Demand features are extracted from the original text data, including explicitness features, completeness features, and compliance features; Based on the key indicators of the project, a comprehensive score label is generated for the historical bidding document dataset; Using the aforementioned demand features as input and the aforementioned comprehensive rating labels as training targets, a machine learning model is trained. Verify and output the trained full-process node model of bidding.

3. The intelligent decision support system for the entire bidding process according to claim 2, characterized in that, The model building module is used for: The required features include explicitness feature extraction, completeness feature extraction, and compliance feature extraction; The explicit feature extraction includes using a natural language processing model to calculate the perplexity of the requirement description text, and / or identifying and counting the frequency of ambiguous words in the text; The integrity feature extraction includes calculating the coverage of the original text data to the feature list based on a predefined feature list, using text matching or entity recognition technology. The compliance feature extraction includes comparing the semantic similarity of the original text data with related texts to identify clauses with compliance risks.

4. The intelligent decision support system for the entire bidding process according to claim 1, characterized in that, The first calculation module is used for: Determine the range of node evaluation values ​​corresponding to the set of node evaluation values, wherein the range of node evaluation values ​​includes a first preset node evaluation value and a second preset node evaluation value, and the first preset node evaluation value is less than the second preset node evaluation value; The set of node evaluation values ​​is traversed. When a node evaluation value is less than the first preset node evaluation value, the difference between the first preset node evaluation value and the node evaluation value is calculated and used as a node evaluation fluctuation metric. When the node evaluation value is greater than the second preset node evaluation value, the difference between the node evaluation value and the second preset node evaluation value is calculated and used as the node evaluation fluctuation metric. When the node evaluation value is within the range of the node evaluation value, the first difference between the node evaluation value and the first preset node evaluation value is calculated, the second difference between the second preset node evaluation value and the node evaluation value is calculated, and the average of the first difference and the second difference is used as the node evaluation fluctuation metric.

5. The intelligent decision support system for the entire bidding process according to claim 1, characterized in that, The second calculation module is used for: The fluctuation metric values ​​of every two nodes are combined to obtain multiple sets of node fluctuation metric values, and the relative fluctuation difference is calculated based on the set of node fluctuation metric values. The bidding difference coefficient for the entire bidding process is calculated based on the relative fluctuation difference.

6. The intelligent decision support system for the entire bidding process according to claim 5, characterized in that, The second calculation module is used for: Calculate the first difference of the node evaluation volatility metric group, wherein the first difference is the absolute value of the difference between the two node evaluation volatility metrics in the node evaluation volatility metric group. Extract the maximum node evaluation volatility metric and the minimum node evaluation volatility metric, and calculate the second difference between the maximum node evaluation volatility metric and the minimum node evaluation volatility metric; Calculate the ratio of the second difference to each of the first differences, and use it as the relative fluctuation difference.

7. The intelligent decision support system for the entire bidding process according to claim 5, characterized in that, The second calculation module is used for: Extract the same relative fluctuation difference from the relative fluctuation difference, and calculate the stability difference coefficient of the entire bidding process based on the same relative fluctuation difference; Extract the remaining relative fluctuation difference from the relative fluctuation difference, and calculate the fluctuation difference coefficient of the entire bidding process based on the remaining relative fluctuation difference; The weighted sum of the stability difference coefficient and the wave state difference coefficient yields the bidding difference coefficient for the entire bidding process.

8. The intelligent decision support system for the entire bidding process according to claim 7, characterized in that, The second calculation module is used for: Based on the same relative volatility difference, multiple relative volatility difference sequences are obtained; The number of first relative volatility difference sequences in the statistical relative volatility difference sequence; Extract one relative volatility difference from each of the relative volatility difference sequences and calculate the sum of the first relative volatility differences; Calculate the mean of relative volatility differences for all identical relative volatility differences, remove all relative volatility difference sequences that are less than the mean of relative volatility differences, and count the number of second relative volatility difference sequences among the remaining relative volatility difference sequences. Extract one relative volatility difference from each of the remaining relative volatility difference sequences, and calculate the second relative volatility difference sum value; The stability difference coefficient for the entire bidding process is calculated based on the number of the first relative fluctuation difference sequence, the number of the second relative fluctuation difference sequence, the sum of the first relative fluctuation difference, and the sum of the second relative fluctuation difference. Where s is the stability difference coefficient of the entire bidding process, r1 is the number of the first relative fluctuation difference sequence, r2 is the number of the second relative fluctuation difference sequence, t1 is the sum of the first relative fluctuation difference, and t2 is the sum of the second relative fluctuation difference.

9. The intelligent decision support system for the entire bidding process according to claim 7, characterized in that, The second calculation module is used for: The waveform difference coefficient for the entire bidding process is calculated using the following formula: Where w is the volatility difference coefficient of the entire bidding process, n is the number of remaining relative volatility differences, and g i For the relative fluctuation difference of the i-th remaining value, g i+1 This represents the relative fluctuation difference of the (i+1)th remaining variable.

10. The intelligent decision support system for the entire bidding process according to claim 1, characterized in that, The process decision module is used for: When the bidding difference coefficient is less than the preset bidding difference coefficient, it is determined that the entire bidding process meets the bidding conditions. When the bidding difference coefficient is greater than or equal to the preset bidding difference coefficient, it is determined that the entire bidding process does not meet the bidding conditions.

Images