Bid evaluation data intelligent processing method and system

Abstract

The application discloses a bid evaluation data intelligent processing method and system, relates to the technical field of electronic bidding and tendering, and comprises the following steps: acquiring multi-source heterogeneous data, performing structured conversion processing on the multi-source heterogeneous data; constructing data based on a model, extracting bid subject features, bid content features and bid behavior features, analyzing multi-layer correlation relationships among the features, and constructing a correlation analysis model; inputting current bid evaluation data into the correlation analysis model, performing abnormal bid identification, and outputting abnormal behavior early warning information; intelligently analyzing and processing the current bid evaluation data, and extracting bid evaluation auxiliary information; generating bid evaluation decision suggestions and constructing a visual interface based on the abnormal behavior early warning information and the bid evaluation auxiliary information; and encrypting and notarizing bid evaluation process data through a distributed storage technology. Through structured processing and feature extraction on multi-source heterogeneous data, the application realizes all-round correlation analysis on bid data, and effectively improves the scientificity and accuracy of bid evaluation decisions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic bidding and tendering, and in particular to a method and system for intelligent processing of bid evaluation data. BACKGROUND

[0002] With the popularization and application of electronic bidding and tendering, the amount of data generated in the bid evaluation process is increasingly large and complex, but the existing bid evaluation system still has obvious deficiencies. The current bid evaluation system mainly relies on manual review or single technical means (such as electronic signature), and lacks the ability to integrate and process multi-source heterogeneous data. In actual operation, the bid party historical data, market dynamics, enterprise credit information and other information are often scattered in different systems, making it difficult to realize multi-dimensional correlation analysis and unable to provide comprehensive decision support for bid evaluation experts.

[0003] In addition, the transparency of the traditional bid evaluation process is insufficient, the bid evaluation records may be tampered with, the subjective inclination of experts is difficult to effectively avoid, and the irregularities such as bid rigging and bid colluding have strong concealment. Bid evaluation experts need to manually read a large number of bid documents during the review process, which is time-consuming and inefficient, especially in the remote bid evaluation scenario, there is a lack of effective expert collaboration mechanism and real scene interaction capability, and the emerging technologies such as blockchain, artificial intelligence and virtual reality in the existing system are often isolated from each other, and have not formed an effective technology integration system. SUMMARY

[0004] The present application provides a method and system for intelligent processing of bid evaluation data to solve the technical problems of insufficient data integration, process opacity and low review efficiency in the existing bid evaluation system.

[0005] Therefore, the first aspect of the present application provides a method for intelligent processing of bid evaluation data, comprising:

[0006] Obtaining multi-source heterogeneous data related to bid evaluation and performing structured conversion processing on the multi-source heterogeneous data; the multi-source heterogeneous data includes model construction data and current bid evaluation data;

[0007] Based on the model construction data, extracting bid subject features, bid content features and bid behavior features, and analyzing the multi-layer correlation between the features to construct a correlation analysis model;

[0008] Inputting the current bid evaluation data into the correlation analysis model, performing abnormal bid identification and outputting abnormal behavior warning information;

[0009] Intelligently analyzing and processing the current bid evaluation data to extract bid evaluation auxiliary information;

[0010] Based on the abnormal behavior warning information and the bid evaluation auxiliary information, generating bid evaluation decision suggestions and constructing a visual interface;

[0011] The bid evaluation process data is encrypted and stored by a distributed storage technology.

[0012] Optionally, the model construction data includes historical bid data, extended verification data and process evidence data; and the current bid-to-evaluate data includes bid party qualification documents, structured technical bid documents and standardized commercial offer documents.

[0013] Optionally, the analysis of the multi-layer correlation between the features includes:

[0014] The correlation index between the bid subject features, bid content features and bid behavior features is calculated, and the connection relationship between the features is determined based on a preset correlation threshold to generate a feature correlation network.

[0015] The nodes in the feature correlation network are classified according to the bid subject features, bid content features and bid behavior features to construct a multi-layer feature network structure.

[0016] The multi-layer feature network structure is topologically analyzed to determine the key nodes and connection paths in the network.

[0017] Using graph embedding technology, the nodes and their topological relationships in the multi-layer feature network structure are converted into low-dimensional vector representations, and high-order correlation patterns between the features are extracted.

[0018] According to the low-dimensional vector representation and the high-order correlation pattern, an association analysis model is established.

[0019] Optionally, the execution of abnormal bid identification and output of abnormal behavior warning information includes:

[0020] An abnormal bid behavior feature library is constructed.

[0021] Based on the association analysis model, the current bid-to-evaluate data is converted into an associated feature vector, and a high-order correlation pattern is extracted.

[0022] The similarity between the associated feature vector and each abnormal type feature vector in the abnormal bid behavior feature library is calculated, and abnormal bid candidates are selected according to a preset similarity threshold.

[0023] Using the high-order correlation pattern, the abnormal bid candidates are clustered and analyzed to generate multiple abnormal bid clusters.

[0024] Abnormal correlation patterns are extracted from the abnormal bid clusters, and the abnormal risk degree is evaluated based on the abnormal correlation patterns.

[0025] Abnormal behavior warning information is generated according to the abnormal risk degree.

[0026] Optionally, the intelligent analysis and processing of the current bid-to-evaluate data includes:

[0027] Analyzing the qualification documents of the bidding party, verifying compliance and validity, and generating qualification evaluation results;

[0028] Analyzing the structured technical tender, evaluating the feasibility, innovation and response degree of the technical content, and forming the technical scoring basis;

[0029] Analyzing the standardized commercial offer document, judging the reasonableness and completeness of the offer, and generating the offer evaluation conclusion;

[0030] Comparative analysis of the differences and advantages and disadvantages of the technical content of different bidding parties, forming horizontal comparison data;

[0031] Integrating the qualification evaluation results, technical scoring basis, offer evaluation conclusion and horizontal comparison data to generate structured evaluation assistance information.

[0032] Optionally, generating evaluation decision suggestions and building a visual interface includes:

[0033] Building a decision matrix, cross-analyzing abnormal behavior warning information and evaluation assistance information, and forming a decision support data structure;

[0034] According to the decision support data structure, generate evaluation decision suggestions including recommended scoring interval, abnormal risk warning mark and evaluation focus prompt;

[0035] Building a multi-level visual interface according to the qualification, technology and business three dimensions, and marking risk points and decision basis in the visual interface;

[0036] Building an interactive scoring input interface, and generating real-time rationality prompt information according to the deviation degree of the expert input score and the system recommended scoring interval;

[0037] Recording the scoring data input by the evaluation expert through the interactive scoring input interface as the expert actual scoring data, and recording the scoring track and modification history to form the evaluation behavior data.

[0038] Optionally, the evaluation process data is encrypted and stored by distributed storage technology, including:

[0039] Packing the evaluation process data and generating data integrity check value; the evaluation process data includes evaluation decision suggestions, expert actual scoring data and evaluation behavior data;

[0040] Using a multi-node distributed storage architecture to encrypt and store the packaged evaluation process data, each node stores different data fragments and saves the corresponding check value;

[0041] Using a threshold signature mechanism to authorize and control the storage nodes.

[0042] The second aspect of the present application provides an intelligent evaluation data processing system, comprising:

[0043] A data preprocessing module is configured to acquire multi-source heterogeneous data related to evaluation and perform structured conversion processing on the multi-source heterogeneous data;

[0044] An association analysis module is configured to construct data based on a model, extract bid subject features, bid content features and bid behavior features, analyze multi-layer association relationships between the features, and construct an association analysis model;

[0045] An abnormal bid identification module is configured to input current to-be-evaluated data into the association analysis model, perform abnormal bid identification, and output abnormal behavior warning information;

[0046] An evaluation assistance analysis module is configured to perform intelligent analysis and processing on the current to-be-evaluated data, and extract evaluation assistance information;

[0047] An evaluation decision support module is configured to generate evaluation decision suggestions and construct a visual interface to support expert evaluation decision-making based on the abnormal behavior warning information and the evaluation assistance information;

[0048] A data encryption and storage module is configured to encrypt and store evaluation process data by using a distributed storage technology.

[0049] The third aspect of the embodiment of the present application provides a computer device, comprising a memory, a processor and a computer program, the computer program is stored in the memory, and the processor runs the computer program to execute the method of the first aspect of the present application and various possible aspects related to the method.

[0050] The fourth aspect of the embodiment of the present application provides a readable storage medium, the readable storage medium stores a computer program, and the computer program is executed by a processor to realize the method of the first aspect of the present application and various possible aspects related to the method.

[0051] The present application has the advantages that: through the structured processing and feature extraction of multi-source heterogeneous data, the present application realizes the all-round association analysis of bid data, effectively improves the scientificity and accuracy of evaluation decision-making; through the calculation of correlation feature vector similarity and high-order correlation pattern clustering analysis, the present application realizes the intelligent identification and risk warning of abnormal bid behaviors such as bid rigging and bid stringing; an evaluation decision support system based on a decision matrix is designed, the abnormal behavior warning information and the evaluation assistance information are cross-analyzed, and a multi-level visual interface is constructed, which significantly improves the transparency and evaluation efficiency of the evaluation process; the distributed storage architecture and the threshold signature mechanism are adopted to perform sharding and encryption storage on the evaluation process data, combined with data integrity verification, the security and non-tamperability of the evaluation data are ensured, and the problem of insufficient data credibility in the traditional evaluation system is fundamentally solved. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0053] Fig. 1 Flow chart of the intelligent evaluation data processing method.

[0054] Fig. 2 Flow chart of the correlation analysis model construction of the intelligent evaluation data processing method.

[0055] Fig. 3 Flow chart of the evaluation decision support and visual interaction of the intelligent evaluation data processing method. DETAILED DESCRIPTION

[0056] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0057] The terms "first", "second", "third", "fourth" and the like (if any) in the description, claims and drawings of the present application are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0058] It should be understood that in various embodiments of the present application, the magnitude of the serial number of each process does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0059] It should be understood that in the present application, "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0060] It should be understood that in the present application, "multiple" means two or more. "And / or" is only a description of the association between the associated objects, which means that there can be three relationships, for example, A and / or B can represent the three cases of A alone, A and B together, and B alone. The character " / " generally represents that the associated objects before and after are in an "or" relationship. "Including A, B and C", "including A, B, C" means that A, B and C are all included, "including A, B or C" means that one of A, B and C is included, and "including A, B and / or C" means that any one or any two or three of A, B and C is included.

[0061] It should be understood that in the present application, "B corresponding to A", "B corresponding to A", "A corresponding to B" or "B corresponding to A" means that B is associated with A, and B can be determined according to A. Determining B according to A does not mean that B is determined only according to A, but also can be determined according to A and / or other information. The matching of A and B means that the similarity of A and B is greater than or equal to a preset threshold.

[0062] Depending on the context, "if" as used herein can be interpreted as "when" or "when" or "in response to determining" or "in response to detecting".

[0063] The technical solutions of the present application will be described in detail below with specific examples. The following specific examples can be combined with each other, and the same or similar concepts or processes can not be described in some examples.

[0064] Example 1, reference Figs. 1-3 , the first embodiment of the present application, the embodiment provides an intelligent evaluation data processing method, the flow chart of the method is as shown in Fig. 1 , the method comprises,

[0065] S1: Obtain multi-source heterogeneous data related to evaluation, and perform structured conversion processing on the multi-source heterogeneous data.

[0066] Among them, the multi-source heterogeneous data includes model construction data and current evaluation data; the model construction data includes historical bidding data, extended verification data and process evidence data; the current evaluation data includes the qualification file of the bidder, the structured technical tender, and the standardized commercial offer file. In addition, the historical bidding data includes authorized historical bidding records and abnormal marking information (such as marking of surrounding bidding, abnormal bidding, and inconsistent qualification), and the historical bidding records include historical qualification files, historical technical tender and historical commercial offer files; the extended verification data includes enterprise credit information and associated enterprise analysis data; the process evidence data includes bidding file submission log, expert electronic signature record and block chain evidence hash value.

[0067] It should be noted that the acquisition process of multi-source heterogeneous data includes: obtaining historical bidding data from the historical bidding library through a secure data interface (two-way authentication), periodic incremental synchronization; receiving the original file uploaded by the bidder through the electronic bidding and tendering system, and generating the current to-be-evaluated bidding data through the pre-defined template analysis; obtaining enterprise credit information including enterprise registration information and administrative penalty records through the API of the national enterprise credit information public system; obtaining associated enterprise equity information including equity structure and actual controller information through a legally authorized data interface; the bidding file submission log contains the de-identified IP address and timestamp; the review expert electronic signature record is generated using a compliant CA digital certificate; the blockchain storage hash value is generated by the blockchain node through cryptographic hash calculation on the core clauses of the bid document.

[0068] Further, the following structured conversion is performed on the multi-source heterogeneous data, including: for unstructured data (such as PDF qualification files), key fields are extracted through OCR technology and converted to JSON format; for semi-structured data (such as XML technical bid documents), check according to the pre-defined XSD template and convert to a relational table; for standardized data (such as CSV business quotes), forcibly unify the currency unit, and check the logical consistency (such as the error between the sub-item quotes and the total price ≤0.01%).

[0069] S2: Based on the model construction data, extract the bidding subject features, bidding content features and bidding behavior features, and analyze the multi-layer association relationship between the features to construct an association analysis model.

[0070] In one specific embodiment of the present application, the association analysis model construction flow chart is as shown in Fig. 2 , including:

[0071] S2.1: Based on the model construction data, extract the bidding subject features, bidding content features and bidding behavior features.

[0072] Specifically, the bidding subject features are extracted, including: extracting the registered capital, qualification grade, historical bid-winning rate and qualification inconsistency markers in abnormal markers of the bidding enterprise from the historical bidding data; extracting the equity structure, actual controller information, associated enterprise list, administrative penalty records from the extended verification data; standardizing the extracted bidding subject features, and encoding the category type features.

[0073] Further, the bidding content features are extracted, including: extracting the technical scheme keywords and key technical parameters (such as construction period, equipment specifications) from the historical technical bid documents; extracting the total price to budget ratio and sub-item price dispersion from the historical business quotation files; extracting the quotation abnormal markers and technical scheme duplication markers from the abnormal marker information; vectorizing the extracted bidding content features, and calculating the feature importance weight.

[0074] Further, the bidding behavior features are extracted, including: extracting IP address geographical distribution, file upload timestamp sequence, modification frequency from the bidding file submission log; extracting review time node, key review opinion change record from the review expert electronic signature record; extracting file hash value from the blockchain storage hash value; extracting around the bidding string label from the abnormal label information; the extracted bidding behavior features are numerically represented, and the time sequence features are time sequence pattern extracted.

[0075] Finally, the dimensions of various features are reduced and the abnormal values are processed to ensure the quality of the features.

[0076] S2.2: Calculate the correlation index between the bidding subject features, bidding content features and bidding behavior features, and determine the connection relationship between the features based on the preset correlation threshold to generate a feature correlation network.

[0077] Specifically, a feature pair set is constructed, and specific features in the bidding subject features, bidding content features and bidding behavior features are paired two by two; according to the feature type, the correlation index of the feature pair is calculated, the Pearson correlation coefficient is calculated for the numerical feature pair, the mutual information value is calculated for the category type feature pair, and the point two column correlation value is calculated for the mixed type feature pair; based on the historical abnormal bidding case data, the optimal correlation threshold is determined through cross-validation as the preset correlation threshold; the correlation index of the feature pair is compared with the preset correlation threshold, and when the correlation index is greater than or equal to the preset threshold, a connection relationship is established between the corresponding features; a feature correlation network is constructed, in which the features are used as network nodes, the connection relationship is used as network edges, and the correlation index value is used as edge weight.

[0078] S2.3: Classify the nodes in the feature correlation network according to the bidding subject features, bidding content features and bidding behavior features, and construct a multi-layer feature network structure.

[0079] Further, each node in the feature correlation network is added with a category identifier, which is marked as a bidding subject feature node, a bidding content feature node or a bidding behavior feature node; three independent feature layers are created, corresponding to the bidding subject feature layer, the bidding content feature layer and the bidding behavior feature layer respectively; the nodes are mapped to the corresponding feature layers according to the category identifier, while the connection relationship between the nodes is preserved; the intra-layer connection and the inter-layer connection are distinguished, forming a multi-layer feature network structure containing three feature layers and their connection relationship.

[0080] Among them, the connection relationship between the nodes in the same feature layer in the feature correlation network is preserved as the intra-layer connection, and the connection relationship between the nodes in different feature layers in the feature correlation network is extracted as the inter-layer connection; a multi-layer feature network structure containing three feature layers and their intra-layer connection and inter-layer connection is constructed.

[0081] S2.4: Topological analysis of the multi-layer feature network structure is performed to determine the key nodes and connection paths in the network.

[0082] Further, the degree centrality of each node in the multi-layer feature network structure is calculated, and nodes that have a connection relationship with multiple other nodes are identified as potential key nodes; the betweenness centrality of each node is calculated, and nodes located on multiple shortest paths are identified as bridge nodes; the eigenvector centrality of each node is calculated, and nodes with greater influence in the overall structure of the network are identified as core nodes; based on the comprehensive score of the three centralities, a set of key nodes in the network is determined; the shortest connection paths between the key nodes are determined, and the path weight is calculated, which is equal to the product of the weights of all edges on the path; the connection paths are sorted based on the path weight, and the top N paths with the highest weight are selected as the key connection paths; inter-layer propagation analysis is performed on the key connection paths to identify abnormal feature propagation chains from the bid subject feature layer to the bid content feature layer to the bid behavior feature layer.

[0083] S2.5: Using graph embedding technology, the nodes and their topological relationships in the multi-layer feature network structure are converted into low-dimensional vector representations, and high-order association patterns between features are extracted.

[0084] Further, an adjacency matrix of the multi-layer feature network structure is constructed to record the connection relationships and weight information between nodes; a random walk strategy is applied to the adjacency matrix to generate node co-occurrence sequences; based on the generated node co-occurrence sequences, a distributed representation learning method is used to train node vector representations, mapping each feature node to a vector space of a predetermined dimension; a hierarchical information weight factor is introduced to weight and adjust the node vectors between different feature layers, enhancing the expression ability of inter-layer associations; the similarity between nodes in the vector space is calculated to construct a node vector similarity matrix; based on the similarity matrix, a clustering algorithm is used to group similar nodes to identify feature combinations that cross layers; subgraph pattern mining methods are used to extract repeated connection patterns from the multi-layer feature network structure; the extracted high-order association patterns are associated with historical abnormal bidding label data to filter out association patterns highly related to abnormal bidding behavior, and an association pattern library is established.

[0085] S2.6: Based on the low-dimensional vector representation and high-order association patterns, an association analysis model is established.

[0086] Preferably, by organizing the bidding subject, content and behavior characteristics into a multi-layer network structure, the correlation analysis model can capture cross-dimensional abnormal propagation paths that are difficult to discover by traditional methods; a variety of centrality indicators are used to identify key nodes and their propagation links, revealing potential collusion relationships; the introduction of hierarchical information weight factors into the graph embedding technology realizes efficient vectorization representation while preserving the topological relationship, significantly improving the identification ability of high-order correlation patterns. These innovations enable the correlation analysis model to discover deep bidding abnormal patterns from complex heterogeneous data, significantly enhancing the identification accuracy of hidden collusion bidding behavior while reducing the false positive rate, providing a more accurate risk warning tool for regulatory authorities.

[0087] S3: input the current bidding data to be evaluated into the correlation analysis model, perform abnormal bidding identification, and output abnormal behavior warning information.

[0088] In one specific embodiment of the present application, step S3 comprises:

[0089] S3.1: build an abnormal bidding behavior feature library.

[0090] Specifically, labeled abnormal bidding cases are extracted from historical bidding data in the model construction data; a multi-layer feature network structure is applied to each abnormal case for feature extraction, and a graph embedding technology is used to convert the features into low-dimensional vector representations; the feature vectors are classified and stored according to the type of abnormality (such as collusion bidding, bid rigging, qualification attachment, etc.); the center point of each type of abnormality feature vector is calculated as the typical feature representation of that type of abnormality, forming a complete abnormal bidding behavior feature library.

[0091] S3.2: based on the correlation analysis model, convert the current bidding data to be evaluated into a correlation feature vector, and extract high-order correlation patterns.

[0092] Further, the current bidding data to be evaluated is received, preprocessed and standardized according to the input format requirements of the correlation analysis model; the preprocessed current bidding data to be evaluated is input into the correlation analysis model to obtain a correlation feature vector; the correlation feature vector is normalized to eliminate dimensional differences; the latent graph structure representing the relationship between features is extracted from the correlation feature vector through forward propagation calculation of the correlation analysis model; the latent correlation structure is matched with the correlation pattern library stored in the model, and the structures with a similarity exceeding the pattern recognition threshold are matched as candidate correlation patterns; the attention mechanism is used to score the importance of the candidate correlation patterns, and the patterns with a score exceeding the importance screening threshold are selected; the selected multiple correlation patterns are combined to form a complete high-order correlation pattern representation.

[0093] Preferably, through the innovative combination of graph embedding technology and attention mechanism, the automatic identification and importance quantification of complex correlation patterns between multi-dimensional features of bid data are realized, effectively improving the detection accuracy of hidden collusive bidding behavior and reducing the dependence of the system on expert experience.

[0094] S3.3: Calculate the similarity between the associated feature vector and each abnormal type feature vector in the abnormal bidding behavior feature library, and filter the abnormal bidding candidates according to the preset similarity threshold.

[0095] Specifically, a hierarchical screening strategy is adopted. First, the associated feature vector of the current bid data to be evaluated and each abnormal type feature vector in the abnormal bidding behavior feature library are subjected to L2 norm normalization, converted into unit vector form; the cosine similarity is used as the preliminary screening index, and the cosine similarity value between the associated feature vector and each abnormal type feature vector is calculated.

[0096] Further, the preliminary screening threshold is set, and the samples with a cosine similarity value lower than the preliminary screening threshold are directly determined as normal bidding behavior without further calculation; for the candidates with a cosine similarity value higher than or equal to the preliminary screening threshold, fine screening is performed.

[0097] In the fine screening stage, according to the data distribution characteristics of the abnormal types, Mahalanobis distance calculation is selectively applied, and the distance value is mapped to a similarity index in the [0, 1] interval through exponential conversion; based on the system's pre-stored historical identification accuracy rate data of abnormal types, it is dynamically determined whether to fuse multiple similarity indexes: for abnormal types with high identification accuracy, only a single optimal index can be used; for difficult-to-identify abnormal types, a weighted fusion strategy is adopted to calculate the comprehensive similarity.

[0098] For the bidding samples with determined similarity, according to the system's pre-defined severity and historical frequency of abnormal types, a set of differentiated similarity thresholds is set for different abnormal types; for each candidate, the top k abnormal types with the highest similarity are determined, where the value of k is dynamically set according to the system resource capacity and processing priority, and only these k types are subjected to threshold judgment; when the similarity is greater than or equal to the corresponding threshold, the bid is marked as a candidate of the corresponding abnormal type; only the candidates determined to be abnormal are sorted according to the similarity from high to low to form a set of abnormal bidding candidates in priority order.

[0099] Preferably, through the innovative combination of hierarchical screening and adaptive similarity fusion mechanism, accurate identification and risk grading of different types of bidding abnormal behavior are realized, greatly improving the system processing efficiency and reducing the false positive rate, especially in complex bidding scenarios with uneven data distribution.

[0100] S3.4: Use high-order association patterns to perform cluster analysis on abnormal bidding candidates and generate multiple abnormal bidding clusters.

[0101] Specifically, the extracted high-order correlation patterns are converted into structured feature vectors of abnormal bidding candidates. For standard features, their activation strength is directly quantified. For complex graph-structured correlation patterns, a graph kernel method is applied to map the topological relationships into a comparable feature space. Based on this feature representation, a similarity matrix is ​​calculated between abnormal bidding candidates. To address the multidimensional nonlinear nature of the correlation patterns, a cosine similarity metric based on the weighted importance of the correlation patterns is employed. Based on the feature dimensionality and data distribution characteristics, clustering algorithms suitable for high-dimensional data (such as K-means or hierarchical clustering) are selected to group abnormal bidding candidates. Clustering parameters are optimized to enhance sensitivity to key pattern differences. Using multiple evaluation metrics such as the silhouette coefficient and Davies-Bouldin index, combined with cross-validation, the optimal number of clusters is determined, ultimately generating multiple internally consistent abnormal bidding clusters.

[0102] S3.5: Extract abnormal correlation patterns from abnormal bidding clusters and evaluate the abnormal risk level based on the abnormal correlation patterns.

[0103] Furthermore, for each abnormal bidding cluster, the common behavioral characteristics among bidders, including price trends, document similarity, and temporal correlation, are analyzed to identify the interaction network structure among bidders within the cluster. Based on the extracted features and interaction structure, a feature representation of abnormal correlation is constructed, quantifying the distribution strength of features across each dimension to form an operationalized abnormal correlation pattern. By comparing the features with historical abnormal case studies, the specific abnormality type corresponding to each cluster (such as collusion, bid rigging, and qualification fraud) is determined, and all associated bidders within the cluster are identified. A comprehensive risk score is calculated based on the strength and complexity of the abnormal correlation pattern, cluster size, and historical risk distribution. This score is then used to classify abnormal bidding clusters and determine the risk level of each cluster.

[0104] S3.6: Generate abnormal behavior warning information based on the degree of abnormal risk.

[0105] Furthermore, based on the risk assessment results of abnormal bidding clusters, a multi-level anomaly warning knowledge base was constructed, mapping risk levels, anomaly types, and associated bidding entity information into structured warning templates. Warning information containing risk identification, anomaly type, and associated entities was generated for each abnormal bidding cluster. High-risk anomalies were automatically linked to relevant evidence chains and behavioral feature descriptions. By extracting multidimensional anomaly association patterns from abnormal bidding clusters and implementing risk quantification assessments, combined with an intelligent warning generation mechanism based on the warning knowledge base and structured templates, the accuracy of abnormal bidding behavior identification and the timeliness of warnings were significantly improved.

[0106] S4: Intelligent analysis and processing of the current bid evaluation data to extract bid evaluation auxiliary information.

[0107] In one specific embodiment of the application, step S4 comprises:

[0108] S4.1: Analyzing the bid party qualification documents, verifying compliance and validity, and generating a qualification evaluation result.

[0109] Specifically, a qualification evaluation rule set is constructed based on project requirements, with necessary conditions and priority conditions being specified; information extraction techniques are used to obtain key information in the qualification documents, including qualification certificate number, level, validity period, and registered personnel information; the extracted qualification information is compared and verified with authoritative databases to confirm the authenticity of the certificate; the historical qualification records of the bid party in the authoritative database are searched, and a time series analysis method is applied to evaluate the stability and development trend of the qualification; the qualification is scored according to the evaluation rule set, and a qualification compliance index is generated; the verification results, stability analysis, and compliance score are integrated to form a complete qualification evaluation result.

[0110] S4.2: Analyzing the structured technical tender, evaluating the feasibility, innovation, and response degree of the technical content, and forming the basis for technical scoring.

[0111] Specifically, a technical evaluation framework and index system are constructed based on the bidding requirements; natural language processing and text mining techniques are applied to automatically extract key technical elements from the technical tender, including technical route, process flow, quality control measures, and resource allocation scheme; the feasibility of the technical scheme is evaluated, and the rationality of technical parameters and the completeness of implementation conditions are analyzed; the innovation points and characteristic contents in the technical scheme are identified, and their application value and technical advancement are evaluated; the content of the technical tender is analyzed for semantic matching with the bidding requirements, and the technical response degree and coverage are quantitatively evaluated through keyword weight and theme correlation degree calculation; the advantages and risk points of the technical scheme are evaluated, and combined with the scoring standard, a systematic technical scoring basis is formed.

[0112] S4.3: Analyzing the standardized commercial offer file, judging the reasonableness and completeness of the offer, and generating an offer evaluation conclusion.

[0113] Specifically, the bid document format specification and content integrity are checked to confirm all items covering the bidding requirements; the unit price and total price information are extracted through optical character recognition and table parsing technology, and the arithmetic accuracy and logical consistency are verified by using automatic calculation verification tools; the ratio of the bid price to the procurement budget is calculated to assess the overall bid level; the bid structure is analyzed, the proportion of each sub-item is calculated and compared with industry standards; the price distribution of all bidders is analyzed using box plot and Z-score statistical methods to identify abnormal high or low price items that are more than three times the standard deviation; the bid items that deviate from the conventional level are reasonably evaluated; and the above analysis results are integrated to generate bid evaluation conclusions.

[0114] S4.4: Comparative analysis of the differences and advantages and disadvantages of the technical content of different bidders to form horizontal comparison data.

[0115] Specifically, key word extraction algorithm and structured data extraction technology are applied to automatically identify and extract key indicator parameters from the technical proposals of each bidder to construct a unified dimensional technical indicator comparison matrix; text similarity calculation model and semantic clustering technology are applied to analyze the content similarity of different bidders' technical proposals to accurately distinguish between general content and individualized content with differences; the route differences of each bidder in key technical links are identified to evaluate the advantages and disadvantages of different technical paths; the technical innovation points and characteristic contents of each bidder are compared to evaluate their application value in the project; the applicability and implementation effect of each technical proposal are evaluated in combination with the project requirements; and the analysis results are integrated to generate structured horizontal comparison data.

[0116] S4.5: Integrate the qualification evaluation results, technical scoring basis, bid evaluation conclusions, and horizontal comparison data to generate structured bid evaluation auxiliary information.

[0117] Specifically, the weight configuration of each evaluation dimension is determined according to the scoring standards in the bidding document; the evaluation results of different dimensions are standardized to establish a unified and comparable scoring system; the comprehensive score and sub-item score of the bidder are calculated to form a ranking result; abnormal situations and points of attention found in the scoring process are marked to provide key basis and analysis explanation for each score; the scoring results, basis explanation, and abnormal marking are organized into a structured data format to form complete structured bid evaluation auxiliary information.

[0118] Preferably, through the multi-dimensional intelligent analysis framework and deep information extraction technology, key information can be automatically identified and extracted from unstructured and semi-structured documents, greatly improving the efficiency and accuracy of bid evaluation data processing; the introduced standardized evaluation method system solves the technical problem that different dimensional evaluation results cannot be directly compared, making the multi-dimensional evaluation results comparable; the innovative integration of abnormal situation identification and labeling function provides technical support for risk warning in the bid evaluation process, reducing the risk of bid evaluation; the key step record mechanism of the scoring process is established to ensure the traceability and fairness of the bid evaluation decision.

[0119] S5: Based on the abnormal behavior warning information and the bid evaluation auxiliary information, a bid evaluation decision suggestion is generated and a visual interface is constructed to support expert review decision-making.

[0120] In one specific embodiment of the present application, the bid evaluation decision support and visual interaction flowchart is as shown in Fig. 3

[0121] S5.1: Construct a decision matrix to cross-analyze abnormal behavior warning information and bid evaluation auxiliary information to form a decision support data structure.

[0122] Specifically, a two-dimensional decision matrix of bidders and review indicators is created; the qualification evaluation results, technical score basis and bid evaluation conclusions are mapped into the decision matrix according to the weight proportion required by the bid evaluation rules; based on the abnormal behavior warning information, the risk level is marked in the decision matrix and clustered and grouped according to the abnormal type to form the area of expert focus; based on the historical scoring data and the current horizontal comparison results, the expected score interval of the review indicators is calculated, and the index points with significant differences are marked; the technical scheme similarity between bidders is calculated, and special marking is performed in the decision matrix when the similarity exceeds the preset threshold; the above analysis results are integrated to construct a three-dimensional decision support data structure containing the dimensions of bidders, review indicators and abnormal association. Through multi-dimensional abnormal feature fusion and intelligent visualization mapping, the correlation analysis of abnormal behavior and bid evaluation elements is realized, and the scientificity and interpretability of bid evaluation decision-making are improved.

[0123] S5.2: According to the decision support data structure, a bid evaluation decision suggestion containing recommended score interval, abnormal risk warning mark and review focus prompt is generated.

[0124] ​Further, specifically, based on the score expectation value and fluctuation interval in the decision support data structure, a recommended score interval is generated for each evaluation index; for the evaluation index with an abnormal risk, a risk level mark is added beside the recommended score interval, and a risk type description is attached; for the bidder with a technical scheme similarity exceeding a preset threshold, an evaluation focus prompt of the technical scheme similarity is generated; for the bidder with an abnormal bid, a price rationality analysis prompt based on the market average price deviation rate is generated; according to the difference significance analysis result of the evaluation index, a score point prompt is generated to guide the experts to focus on the key difference points; the above analysis results are integrated to comprehensively evaluate the advantages, disadvantages and risks of each bidder, and form a structured evaluation content; the overall recommendation is generated by comprehensively evaluating each dimension, including whether to recommend, the recommended ranking interval and the focus; the generated various recommendations are classified and arranged according to the three dimensions of qualification, technology and business to form a structured evaluation decision suggestion document. The evaluation decision suggestion system realizes intelligent assistance and risk prevention and control in the evaluation process through the deep integration of multi-dimensional abnormal features and professional evaluation indexes, and significantly improves the objectivity and professional level of evaluation decision.

[0125] S5.3: The evaluation decision suggestion is constructed into a multi-level visual interface according to the three dimensions of qualification, technology and business, and the risk points and decision basis are marked in the visual interface.

[0126] Further, a hierarchical visual interface based on the three dimensions of qualification, technology and business is designed, and logical connections are established between layers through associated indexes and risk elements; in the qualification evaluation layer, the qualification evaluation results of each bidder are displayed, and the qualification abnormal points are marked with different marks; in the technology evaluation layer, a technology index radar chart is constructed to display the scores of each bidder on different technology indexes; in the business evaluation layer, a price comparison chart and a price analysis trend chart are constructed to highlight the price abnormal points and market reference price.

[0127] S5.4: An interactive score input interface is constructed for evaluation experts to input scores, and real-time rationality prompt information is generated according to the deviation degree of the score input by the experts and the recommended score interval of the system.

[0128] Further, a score input matrix is constructed, each evaluation expert corresponds to a row, and each evaluation index corresponds to a column; the recommended score interval of the system is displayed in the score input interface and marked with a specific mark; the score value input by the expert is monitored, and the deviation rate from the midpoint of the recommended score interval is calculated; when the deviation rate exceeds a preset threshold, a high deviation warning is triggered, and a rationality prompt that needs to be confirmed by the evaluation expert is generated.

[0129] In the rationality prompt, the analysis of key factors leading to high-deviation judgments is displayed, including historical score distribution, key technical point analysis and abnormal risk prompts; the expert confirmation operations and modification trajectories of high-deviation scores are recorded, including the initial score, number of modifications, final score and confirmation reasons; the score distribution of the review experts is analyzed, and when the scores are polarized, a disagreement reminder is sent to the review team leader; the score consistency index is calculated, and when the consistency index exceeds the preset threshold, the review team leader is advised to organize an expert discussion; an annotation function is provided, allowing review experts to add text descriptions and basis for scores that deviate from the recommended range; a score trace diagram is generated to show the score adjustment process and final results of each expert; in the rationality prompt, the analysis of key factors leading to high-deviation judgments is displayed, including historical score distribution, key technical point analysis and abnormal risk prompts.

[0130] S5.5: Record the scoring data entered by the review experts through the interactive scoring input interface as the actual scoring data of the experts, and record the scoring trajectory and modification history to form the review behavior data.

[0131] Specifically, it records and visualizes the expert scoring trajectory, including the initial score, revision process, and final score, while also providing annotations to allow for the addition of scoring rationale. It also analyzes the consistency of the expert review panel's scores. When significant disagreements arise or the consistency index exceeds a preset threshold, an analysis report is sent to the review team leader with coordination recommendations. Through intelligent scoring monitoring and real-time feedback mechanisms, this interactive scoring system ensures the independent judgment of review experts while providing data support. This ensures standardized guidance and anomaly prevention during the bid evaluation process, effectively improving the fairness and scientific nature of the review results.

[0132] S6: Encrypt and store the bidding process data through distributed storage technology.

[0133] In a specific embodiment of the present invention, step S6 includes:

[0134] S6.1: Package the bid evaluation process data and generate a data integrity check value.

[0135] The bid evaluation process data includes evaluation decision recommendations, actual expert scoring data, and review behavior data. Furthermore, a data integrity check value is generated using the SHA-256 hash algorithm combined with digital signature technology. This involves calculating a SHA-256 hash value on the packaged data; digitally signing the hash value using the bid evaluation system's private key; and combining the original hash value with the digital signature to form a complete data check value, ensuring data integrity and immutability.

[0136] S6.2: Utilize a multi-node distributed storage architecture to encrypt and store the packaged evaluation process data in shards. Each node stores different data shards and saves the corresponding checksum values.

[0137] Specifically, the data encryption key is divided into multiple parts by using a key sharing algorithm, wherein any preset number of key fragments can reconstruct the complete key; the packaged data is divided into qualification review data, technical review data, business review data and comprehensive review data four logical fragments; each logical fragment is encrypted to generate an encrypted data fragment; the check value of each encrypted data fragment is calculated to generate a data integrity check code; a chain storage structure is constructed, each data fragment is taken as a block, including its own data, a timestamp, a previous block check value and a current block check value; the encrypted data fragment and the corresponding check value are deployed to different storage nodes; role-based permission management and operation log recording are realized on each storage node; data access timeliness control is set, authorized personnel are allowed to review the complete data within a preset time period after the bid evaluation ends; a data recovery mechanism is constructed, when historical bid evaluation data needs to be reviewed, the key is reconstructed and the data is decrypted after multi-node authorization verification; an audit tracking mechanism is realized, all data access and operation behaviors are recorded, and the evidence is stored through independent nodes.

[0138] S6.3: A threshold signature mechanism is used to authorize and control the storage nodes, and when more than the minimum authorized node number threshold is obtained, the node signature authorization is allowed to access the packaged bid evaluation process data.

[0139] Among them, the minimum authorized node number threshold is determined by two-thirds of the total number of nodes, and is dynamically adjusted in combination with the security level of the bid evaluation project. For ordinary security level bid evaluation projects, it is set to N / 2+1 (N is the total number of nodes); for high security level bid evaluation projects, it is set to 2N / 3+1; for special security projects, it is set to 3N / 4+1, thereby realizing differentiated authorization control under different security levels.

[0140] In addition, it should be noted that the distributed storage technology realized by the present application adopts a combination of asymmetric encryption and homomorphic encryption, supports statistical analysis and audit query of the bid evaluation data in the encrypted state, and can execute specific data analysis tasks without complete decryption. At the same time, the system realizes a data traceability mechanism based on block chain, which can trace every data change in the whole bid evaluation process, ensures the security, integrity and traceability of the bid evaluation data in the storage process, and effectively prevents the risk of data leakage and tampering caused by internal and external threats.

[0141] Further, the embodiment also provides an intelligent evaluation data processing system, comprising: a data preprocessing module, configured to acquire multi-source heterogeneous data related to evaluation, and perform structured conversion processing on the multi-source heterogeneous data; an association analysis module, configured to construct data based on a model, extract bid subject features, bid content features and bid behavior features, analyze multi-layer association relationships between the features, and construct an association analysis model; an abnormal bid identification module, configured to input current to-be-evaluated data into the association analysis model, perform abnormal bid identification, and output abnormal behavior early warning information; an evaluation decision support module, configured to generate evaluation decision suggestions and construct a visual interface to support expert evaluation decision based on the abnormal behavior early warning information and evaluation auxiliary information; and a data encryption storage module, configured to perform encryption storage on evaluation process data through a distributed storage technology.

[0142] To sum up, the application realizes all-round association analysis on bid data through structured processing and feature extraction on multi-source heterogeneous data, effectively improves the scientificity and accuracy of evaluation decision, realizes intelligent identification and risk early warning on abnormal bid behaviors such as bid rigging through calculation of correlation feature vector similarity and high-order correlation pattern clustering analysis, designs an evaluation decision support system based on a decision matrix, cross-analyzes abnormal behavior early warning information and evaluation auxiliary information, and constructs a multi-level visual interface, which significantly improves the transparency and evaluation efficiency of the evaluation process, and adopts a distributed storage architecture and a threshold signature mechanism to perform sharding encryption storage on evaluation process data, in combination with data integrity verification, to ensure the security and non-tamperability of evaluation data, and fundamentally solves the problem of insufficient data credibility in traditional evaluation systems.

[0143] Embodiment 2, which is the second embodiment of the application, is different from the previous embodiment in that:

[0144] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the application or the part of the technical solutions that essentially contribute to the prior art or the part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device) to execute all or part of the steps of the method described in the embodiments of the application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0145] The logic and / or steps represented in the flow diagrams or otherwise described herein, for example, can be considered as a sequence of executable instructions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber (optical), and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that can be later executed by a computer. In this context, a "computer-readable medium" can be any means that can store the program for use by or in connection with the instruction execution system, apparatus, or device.

[0146] The foregoing description, for purposes of explanation, only is specific to the embodiment shown and described herein. It is not intended to be exhaustive or to be limited to the exact form described herein. Many modifications and variations are possible in light of the foregoing teaching. Such modifications and variations that would yet be encompassed by the spirit and scope of the presently disclosed inventive subject matter might include, for example, employing the teachings of the presently disclosed inventive subject matter in various environments and with various technologies. Accordingly, the patentable scope of the present disclosure is defined by the following claims.

[0147] It should be understood that aspects of the application can be implemented in hardware, software, firmware or a combination thereof. In the above embodiment, the various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or their combinations can be used: discrete logic circuitry having logic gates for implementing logic functions upon an application of data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0148] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing the technical solutions of the present application, not to limit the present application; although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that they can still make modifications to the technical solutions recorded in the above-mentioned embodiments, or make equivalent replacements to some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. The intelligent processing method for bid evaluation data is characterized by: include: Acquiring multi-source heterogeneous data related to bid evaluation and performing structured conversion processing on the multi-source heterogeneous data; the multi-source heterogeneous data includes model building data and current bid evaluation data; Building data based on the model, extracting bidding subject characteristics, bidding content characteristics, and bidding behavior characteristics, and analyzing the multi-layer correlation relationships between the characteristics to build a correlation analysis model; Inputting the current bid evaluation data into the association analysis model, performing abnormal bid identification and outputting abnormal behavior warning information; Performing intelligent analysis and processing on the current bid evaluation data to extract auxiliary bid evaluation information; Based on the abnormal behavior warning information and the bid evaluation auxiliary information, generate bid evaluation decision suggestions and build a visual interface; Encrypt and store bid evaluation process data through distributed storage technology; The multi-layer association relationship between the analysis features and the construction of the association analysis model include: Calculating correlation indices among the bidding subject characteristics, bidding content characteristics, and bidding behavior characteristics, and determining connection relationships among the characteristics based on a preset correlation threshold to generate a feature correlation network; Classifying the nodes in the feature correlation network according to bidding subject characteristics, bidding content characteristics and bidding behavior characteristics to construct a multi-layer feature network structure; Performing topological analysis on the multi-layer characteristic network structure to determine key nodes and connection paths in the network; Using graph embedding technology, the nodes and their topological relationships in the multi-layer feature network structure are converted into low-dimensional vector representations, and high-order correlation patterns between features are extracted; Establishing an association analysis model based on the low-dimensional vector representation and the high-order association pattern; The performing of abnormal bidding identification and outputting abnormal behavior warning information includes: Build a feature library of abnormal bidding behavior; Based on the association analysis model, the current bid evaluation data is converted into an association feature vector, and a high-order association pattern is extracted; Calculating the similarity between the associated feature vector and each abnormal type feature vector in the abnormal bidding behavior feature library, and screening abnormal bidding candidates according to a preset similarity threshold; performing cluster analysis on the abnormal bidding candidates using the high-order association pattern to generate a plurality of abnormal bidding clusters; extracting abnormal correlation patterns from the abnormal bidding clusters, and evaluating the abnormal risk level based on the abnormal correlation patterns; Generate abnormal behavior warning information based on the abnormal risk level.

2. The method for intelligent processing of bid evaluation data according to claim 1, characterized in that: include: The model building data includes historical bidding data, extended verification data and process evidence data; the current data to be evaluated includes bidder qualification documents, structured technical bid documents and standardized business quotation documents.

3. The method for intelligent processing of bid evaluation data according to claim 1, characterized in that: The intelligent analysis and processing of the current bid evaluation data includes: Analyze bidder qualification documents, verify compliance and validity, and generate qualification assessment results; Analyze structured technical bids, evaluate the feasibility, innovation and responsiveness of technical content, and form a basis for technical scoring; Analyze standardized commercial quotation documents, determine the rationality and completeness of quotations, and generate quotation evaluation conclusions; Compare and analyze the differences, strengths and weaknesses of the technical contents of different bidders to form horizontal comparison data; Integrate the qualification assessment results, technical scoring basis, quotation evaluation conclusions and horizontal comparison data to generate structured bid evaluation auxiliary information.

4. The method for intelligent processing of bid evaluation data according to claim 1, characterized in that: Generating the bid evaluation decision suggestion and building the visual interface includes: Constructing a decision matrix, cross-analyzing the abnormal behavior warning information and the bid evaluation auxiliary information to form a decision support data structure; Generate a bid evaluation decision suggestion including a recommended scoring range, abnormal risk warning mark and evaluation key prompts based on the decision support data structure; Construct a multi-level visual interface for the bid evaluation decision suggestions according to the three dimensions of qualifications, technology, and business, and mark the risk points and decision basis in the visual interface; Build an interactive rating input interface and generate real-time rationality prompt information based on the degree of deviation between the expert input rating and the system recommended rating range; The scoring data input by the review experts through the interactive scoring input interface is recorded as the actual scoring data of the experts, and the scoring trajectory and modification history are recorded to form review behavior data.

5. The method for intelligent processing of bid evaluation data according to claim 1, characterized in that: The encrypted storage of bid evaluation process data by distributed storage technology includes: Packaging the bid evaluation process data and generating a data integrity check value; the bid evaluation process data includes bid evaluation decision recommendations, actual expert scoring data, and review behavior data; Using a multi-node distributed storage architecture, the packaged bid evaluation process data is encrypted and stored in shards. Each node stores different data shards and saves the corresponding checksum values. A threshold signature mechanism is used to authorize and control storage nodes.

6. The bid evaluation data intelligent processing system is characterized by: include: A data preprocessing module is used to obtain multi-source heterogeneous data related to bid evaluation and perform structured conversion processing on the multi-source heterogeneous data; The association analysis module is used to construct data based on the model, extract the characteristics of the bidding subject, bidding content and bidding behavior, analyze the multi-layer association relationship between the characteristics, and build an association analysis model; An abnormal bid identification module is used to input the current bid evaluation data into the association analysis model, perform abnormal bid identification and output abnormal behavior warning information; A bid evaluation auxiliary analysis module is used to perform intelligent analysis and processing on the current bid evaluation data and extract bid evaluation auxiliary information; A bid evaluation decision support module is used to generate bid evaluation decision suggestions and build a visual interface to support expert review decisions based on the abnormal behavior warning information and the bid evaluation auxiliary information; Data encryption and evidence storage module, used to encrypt and store bid evaluation process data through distributed storage technology; The multi-layer association relationship between the analysis features and the construction of the association analysis model include: Calculating correlation indices among the bidding subject characteristics, bidding content characteristics, and bidding behavior characteristics, and determining connection relationships among the characteristics based on a preset correlation threshold to generate a feature correlation network; Classifying the nodes in the feature correlation network according to bidding subject characteristics, bidding content characteristics and bidding behavior characteristics to construct a multi-layer feature network structure; Performing topological analysis on the multi-layer characteristic network structure to determine key nodes and connection paths in the network; Using graph embedding technology, the nodes and their topological relationships in the multi-layer feature network structure are converted into low-dimensional vector representations, and high-order correlation patterns between features are extracted; Establishing an association analysis model based on the low-dimensional vector representation and the high-order association pattern; The performing of abnormal bidding identification and outputting abnormal behavior warning information includes: Build a feature library of abnormal bidding behavior; Based on the association analysis model, the current bid evaluation data is converted into an association feature vector, and a high-order association pattern is extracted; Calculating the similarity between the associated feature vector and each abnormal type feature vector in the abnormal bidding behavior feature library, and screening abnormal bidding candidates according to a preset similarity threshold; performing cluster analysis on the abnormal bidding candidates using the high-order association pattern to generate a plurality of abnormal bidding clusters; extracting abnormal correlation patterns from the abnormal bidding clusters, and evaluating the abnormal risk level based on the abnormal correlation patterns; Generate abnormal behavior warning information based on the abnormal risk level.

7. A computer device, characterized in that: include: A memory, a processor, and a computer program, wherein the computer program is stored in the memory, and the processor runs the computer program to execute the method according to any one of claims 1 to 5.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which is used to implement the method according to any one of claims 1 to 5 when executed by a processor.

Images