Data association analysis method and device in bidding and tendering field

Abstract

The invention discloses a data association analysis method and device in the field of bidding and tendering. The method comprises the steps that a bidding and tendering historical information set is acquired; the bidding and tendering historical information set comprises a historical bidding information set and a historical bid winning information set; the historical bid winning information set comprises historical bid winning information; the historical bid-winning information comprises a bid-winning type, a behavior type, a bid-winning scheme technology description statement and a corresponding technology number; performing rule mining processing on the bidding and tendering historical information set to obtain a rule set; the rule set comprises a behavior technology matching model and a behavior pattern sequence pair library; acquiring a bidding document information set; the bidding document information set comprises bidding document information; the bidding document information comprises bidding time, a behavior type, a bidding technical route number and a bidding type; and processing the bidding document information set by utilizing the rule set to obtain an associated information set.

Description

Technical Field

[0001] The present invention relates to the fields of bidding and procurement technology and data processing, and in particular to a data association analysis method and device in the bidding field. Background Art

[0002] In the bidding and tendering sector, with the increasing use of informatization, a vast amount of bidding and tendering data has been accumulated. This data covers a wide range of information, including bidders' bidding behavior, winning bids, and descriptions of various technical proposals. However, the current analysis and utilization of this data still faces many shortcomings. Traditional bidding and tendering data analysis methods primarily focus on examining the surface information of individual bid documents, such as basic elements like bid price and bid time. This analysis is limited in its analysis of bidders' behavioral patterns and the correlations between technical proposals. This approach struggles to uncover potential correlations, such as similarities between different bidders' technical proposals or variations in the behavior of the same bidder across different projects. This can lead to problems, such as the inability to effectively identify potential bid-rigging and collusion, and the inability to provide bidders with a more accurate basis for evaluating technical proposals. Furthermore, it lacks in-depth assessment of bidders' technical strength and development trends, hindering the healthy development of the bidding and tendering market. Effective correlation analysis of bidding and tendering data to extract relevant information on potential bid-rigging and collusion remains a pressing challenge. Summary of the Invention

[0003] The present invention mainly solves the problem of how to conduct effective correlation analysis on bidding data and extract relevant information on potential bid rigging and collusion. The present invention discloses a data correlation analysis method and device in the bidding field.

[0004] In a first aspect, an embodiment of the present invention discloses a data association analysis method in the bidding field, comprising:

[0005] S1, collecting and obtaining a set of historical bidding information; the set of historical bidding information includes a set of historical bidding information and a set of historical winning bid information; the set of historical bidding information includes historical bidding information; the historical bidding information includes a description of bidding behavior; the set of historical winning bid information includes historical winning bid information; the historical winning bid information includes a winning bid type, behavior type, a technical description of the winning bid solution, and a corresponding technical number;

[0006] S2, performing rule mining processing on the bidding history information set to obtain a rule set; the rule set includes a behavior technology matching model and a behavior pattern sequence pair library;

[0007] S3, collecting and obtaining a bidding document information set; the bidding document information set includes bidding document information; the bidding document information includes bidding time, bidding IP address, behavior type, bidding technical route number and bidding type;

[0008] S4: Process the bidding document information set using the rule set to obtain a related information set.

[0009] The rule mining process is performed on the bidding history information set to obtain a rule set, including:

[0010] S21, constructing a first bidding behavior pattern sequence library and a bidding mapping model using the bidding history information set;

[0011] S22, using the bidding mapping model and the first bidding behavior pattern sequence library to construct a second bidding behavior pattern sequence library and a behavior technology matching model;

[0012] S23, using the second bidding behavior pattern sequence library, construct a behavior pattern sequence pair library; the rule set includes a behavior technology matching model and a behavior pattern sequence pair library.

[0013] The method of constructing a first bidding behavior pattern sequence library and a bidding mapping model by utilizing the historical bidding information set and the historical winning bid information set includes:

[0014] S211, performing an extraction operation on each piece of historical bidding information in the historical bidding information set to obtain a description statement of the corresponding bidding behavior;

[0015] S212: sorting the bidding behavior description sentences extracted from each historical bidding information according to the appearance time of the description sentences to obtain a first bidding behavior pattern sequence of the historical bidding information; the first bidding behavior pattern sequence is a directed sequence of bidding behavior description sentences;

[0016] S213, integrating the first bidding behavior pattern sequences of all historical bidding information to obtain a first bidding behavior pattern sequence library;

[0017] S214, performing semantic similarity calculation on the description of the bidding behavior in the first bidding behavior pattern sequence and the technical description of the winning bid solution in the historical winning bid information set to obtain a technical similarity value;

[0018] S215: Determine whether the technical similarity value is greater than a set similarity threshold; if so, establish a mapping relationship between the technical description of the winning bid in the historical winning bid information set and the bidding behavior corresponding to the description of the bidding behavior in the first bidding behavior pattern sequence;

[0019] S216: Utilize all established mapping relationships to construct a bidding mapping model.

[0020] The semantic similarity calculation of the description statement of the bidding behavior of the first bidding behavior pattern sequence and the technical description statement of the winning bid solution in the historical winning bid information set to obtain a technical similarity value includes:

[0021] A first description matrix is ​​constructed using the description statements of all bidding behaviors in the first bidding behavior pattern sequence; a row vector of the first description matrix is ​​a vector of a description statement of a bidding behavior in the first bidding behavior pattern sequence;

[0022] A second description matrix is ​​constructed using all technical description statements of the winning bid solutions in the historical winning bid information set; a row vector of the second description matrix is ​​a vector of a technical description statement of a winning bid solution in the historical winning bid information set;

[0023] Difference calculation is performed on the first description matrix and the second description matrix to obtain a technical similarity value.

[0024] The performing difference calculation on the first description matrix and the second description matrix to obtain a technical similarity value includes:

[0025] Subtracting the first description matrix from the second description matrix to obtain a difference matrix;

[0026] Performing eigenvector calculation on the difference matrix to obtain a first eigenvector and a second eigenvector;

[0027] Performing EEMD transformation on the second eigenvector to obtain a eigentransformation vector;

[0028] A fusion calculation is performed on the first feature vector and the feature transformation vector to obtain a technical similarity value.

[0029] The expression for calculating the eigenvector is:

[0030]

[0031]

[0032] Among them, p1 i and p2 iare the i-th elements of the first and second eigenvectors, respectively, P ij is the element of the i-th row and j-th column of the difference matrix, α i is the mean of the i-th row of the difference matrix, β i is the variance of the i-th row of the difference matrix, λ is the mean of all elements of the difference matrix, and N is the column dimension of the difference matrix;

[0033] The expression of the fusion calculation is:

[0034]

[0035] Among them, yf is the technical similarity value, M is the total number of elements of the first eigenvector, H 2i is the i-th element of the feature transformation vector.

[0036] The method of analyzing and processing the bidding document information set by using the rule set to obtain a related information set includes:

[0037] S31, grouping the bidding document information using the bidding IP address to obtain a plurality of bidding document information sets; a bidding document information set includes a plurality of bidding document information having the same bidding IP address;

[0038] S32, sorting all the bidding document information in each bidding document information set according to their bidding time to obtain a corresponding bidding document information sequence;

[0039] S33, for each bidding document information sequence, performing technical correlation determination on every two adjacent bidding document information in the bidding document information sequence to obtain technical correlation relationship information of the bidding document information sequence;

[0040] S34, performing feature association recognition on the technical association relationship information of all bidding document information sequences to obtain an association information set.

[0041] According to a second aspect of an embodiment of the present invention, a data association analysis device in the field of bidding is disclosed, the device comprising:

[0042] a memory storing executable program code;

[0043] a processor coupled to the memory;

[0044] The processor calls the executable program code stored in the memory to execute the data association analysis method in the bidding field.

[0045] According to a third aspect of an embodiment of the present invention, a computer-storable medium is disclosed, wherein the computer-storable medium stores computer instructions. When the computer instructions are called by a computer, the computer instructions are used to execute the data association analysis method in the bidding field.

[0046] According to a fourth aspect of an embodiment of the present invention, an information data processing terminal is disclosed, which is used to implement the data association analysis method in the bidding field.

[0047] The beneficial effects of the present invention are:

[0048] This invention constructs a behavioral technology matching model and a behavioral pattern sequence pair library by conducting in-depth rule mining on historical bidding information sets. By utilizing these rule sets to process bid document information sets, it is possible to accurately identify technical relationships between bid documents. For example, similarities between different bidders' technical solution descriptions can be detected. Even if these descriptions differ in wording, semantic similarity calculations can still accurately identify them. This helps the tendering party gain a more comprehensive understanding of the bidders' technical strengths and potential cooperative or competitive relationships.

[0049] By analyzing bidding behavior pattern sequences, the present invention can identify abnormal behavior pattern sequence pairs. For example, if multiple bidders frequently exhibit similar bidding behavior patterns across different projects, and these patterns closely match certain patterns in historical bid winning information, but the winning bids do not conform to normal competition rules, this may indicate the presence of unfair practices such as bid rigging or collusion. This invention can provide regulatory authorities with powerful clues to effectively prevent unfair competition in the bidding and tendering sector and maintain fairness and justice in the bidding and tendering market. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 4 is an implementation flow chart of the method of the present invention. DETAILED DESCRIPTION

[0051] In order to better understand the content of the present invention, an embodiment is given here.

[0052] Figure 1 4 is an implementation flow chart of the method of the present invention.

[0053] In a first aspect, an embodiment of the present invention discloses a data association analysis method in the bidding field, comprising:

[0054] S1, collecting and obtaining a set of historical bidding information; the set of historical bidding information includes a set of historical bidding information and a set of historical winning bid information; the set of historical bidding information includes historical bidding information; the historical bidding information includes a description of bidding behavior; the set of historical winning bid information includes historical winning bid information; the historical winning bid information includes a winning bid type, behavior type, a technical description of the winning bid solution, and a corresponding technical number;

[0055] Each piece of information in the historical bidding information set and the historical winning bid information set has a corresponding relationship according to the time of occurrence.

[0056] S2, performing rule mining processing on the bidding history information set to obtain a rule set; the rule set includes a behavior technology matching model and a behavior pattern sequence pair library;

[0057] S3, collecting and obtaining a bidding document information set; the bidding document information set includes bidding document information; the bidding document information includes bidding time, behavior type, bidding IP address, bidding technical route number and bidding type;

[0058] S4: Process the bidding document information set using the rule set to obtain a related information set.

[0059] The implementation of this invention can automate and intelligently analyze complex bidding data. Document review and correlation analysis, which previously required extensive manual effort, can now be rapidly completed using rule sets. This significantly improves the efficiency of bidding management, reduces the time and cost of manual review, and makes the bidding management process more efficient and transparent.

[0060] The rule mining process is performed on the bidding history information set to obtain a rule set, including:

[0061] S21, constructing a first bidding behavior pattern sequence library and a bidding mapping model using the bidding history information set;

[0062] S22, using the bidding mapping model and the first bidding behavior pattern sequence library to construct a second bidding behavior pattern sequence library and a behavior technology matching model;

[0063] S23, using the second bidding behavior pattern sequence library, construct a behavior pattern sequence pair library; the rule set includes a behavior technology matching model and a behavior pattern sequence pair library.

[0064] The method of constructing a first bidding behavior pattern sequence library and a bidding mapping model by utilizing the historical bidding information set and the historical winning bid information set includes:

[0065] S211, performing an extraction operation on each piece of historical bidding information in the historical bidding information set to obtain a description statement of the corresponding bidding behavior;

[0066] S212: sorting the bidding behavior description sentences extracted from each historical bidding information according to the appearance time of the description sentences to obtain a first bidding behavior pattern sequence of the historical bidding information; the first bidding behavior pattern sequence is a directed sequence of bidding behavior description sentences;

[0067] S213, integrating the first bidding behavior pattern sequences of all historical bidding information to obtain a first bidding behavior pattern sequence library;

[0068] S214, performing semantic similarity calculation on the description of the bidding behavior in the first bidding behavior pattern sequence and the technical description of the winning bid solution in the historical winning bid information set to obtain a technical similarity value;

[0069] S215: Determine whether the technical similarity value is greater than a set similarity threshold; if so, establish a mapping relationship between the technical description of the winning bid in the historical winning bid information set and the bidding behavior corresponding to the description of the bidding behavior in the first bidding behavior pattern sequence;

[0070] S216: Utilize all established mapping relationships to construct a bidding mapping model.

[0071] The directed sequence means that the previous element in the sequence points to the next element, and the previous element and the next element have a fixed pointing relationship.

[0072] The extraction operation can be implemented using TextBlob or PyTorch-NLP for natural language processing;

[0073] The construction of bidding behavior pattern sequences is based on the bidding scheme technology and procedural content contained in the bidding behaviors at different steps in historical bidding information. The construction of bidding behavior pattern sequences primarily identifies the execution order relationship between bidding behaviors at different steps in the current bidding event as the basis for association, thereby constructing bidding behavior pattern sequences to describe the bidding behavior process.

[0074] The historical winning bid information set is obtained from the historical winning bid data information set.

[0075] The semantic similarity calculation of the description statement of the bidding behavior of the first bidding behavior pattern sequence and the technical description statement of the winning bid solution in the historical winning bid information set to obtain a technical similarity value includes:

[0076] A first description matrix is ​​constructed using the description statements of all bidding behaviors in the first bidding behavior pattern sequence; a row vector of the first description matrix is ​​a vector of a description statement of a bidding behavior in the first bidding behavior pattern sequence;

[0077] The serial numbers of the row vectors of the first description matrix are determined from small to large according to the order of occurrence of the bidding behavior;

[0078] The vector of the description sentence is obtained by processing the description sentence using the word2vec command or the text vector conversion command in Python.

[0079] A second description matrix is ​​constructed using all technical description statements of the winning bid solutions in the historical winning bid information set; a row vector of the second description matrix is ​​a vector of a technical description statement of a winning bid solution in the historical winning bid information set;

[0080] Calculating the difference between the first description matrix and the second description matrix to obtain a technical similarity value;

[0081] The performing difference calculation on the first description matrix and the second description matrix to obtain a technical similarity value includes:

[0082] Subtracting the first description matrix from the second description matrix to obtain a difference matrix;

[0083] Performing eigenvector calculation on the difference matrix to obtain a first eigenvector and a second eigenvector;

[0084] Performing EEMD transformation on the second eigenvector to obtain a eigentransformation vector;

[0085] Performing a fusion calculation on the first feature vector and the feature transformation vector to obtain a technology similarity value;

[0086] The expression for calculating the eigenvector is:

[0087]

[0088] Among them, p1 i and p2 i are the i-th elements of the first and second eigenvectors, respectively, P ij is the element of the i-th row and j-th column of the difference matrix, α i is the mean of the i-th row of the difference matrix, β i is the variance of the i-th row of the difference matrix, λ is the mean of all elements of the difference matrix, and N is the column dimension of the difference matrix;

[0089] The expression of the fusion calculation is:

[0090]

[0091] Among them, yf is the technical similarity value, M is the total number of elements of the first eigenvector, H 2i is the i-th element of the feature transformation vector.

[0092] The fusion calculation expression performs EEMD transformation on the second eigenvector, decomposing the nonlinear and non-stationary bidding behavior characteristics into intrinsic mode functions (IMFs) of different time scales to generate a characteristic transformation vector. This process can extract the trend characteristics of the bidding behavior sequence (such as the evolution of technical solutions of long-term cooperative bidding groups in multiple rounds of bidding) and suppress short-term random fluctuations. 2i A weighted summation of the logarithmic terms of the first eigenvector amplifies the impact of outliers in the eigenvector (e.g., unusually high similarity between bidding behavior and the winning technology at a certain stage), making the final similarity value more sensitive to characteristic patterns associated with bid rigging and collusion, facilitating the setting of thresholds for rapid screening of suspicious cases. Fusion computing combines the characteristics of bidding behavior sequences with the semantic features of technical solutions, enabling cross-dimensional correlation analysis between "behavioral patterns" and "technical content." For example, when different bidders share the same IP address (likely representing the same group) and the similarity yf between the eigenvectors of their bidding behavior sequences and historically winning technologies consistently exceeds a threshold, a high-risk bid rigging behavior can be comprehensively identified. This mechanism avoids the limitations of single-dimensional analysis (such as analyzing only IP address duplication or technical text similarity) and enhances the accuracy and comprehensiveness of correlation analysis through multi-source data fusion. The numerical value of the similarity yf directly reflects the degree of correlation between bidding behavior and winning technology, making it easier for business personnel to set appropriate thresholds based on historical data (e.g., triggering an alert when yf > 0.8). This enables an interpretable mapping from data features to business rules, lowering the technical barrier to model application.

[0093] The eigenvector calculation, which nonlinearly transforms the elements of the difference matrix using an exponential function (Gaussian kernel), can capture implicit associations between the bidding behavior description and the winning technical description (such as non-directly matching but semantically related features), avoiding the omission of complex associations in linear calculations. Logarithmic and geometric mean operations: Utilizing the logarithmic function to compress the numerical range, combined with the geometric mean (Nth-root product), highlight the overall consistency of features across different dimensions. This is suitable for identifying gradual similarities or abnormal fluctuations in technical descriptions during multi-stage bidding (e.g., bid-rigging rings evade scrutiny by adjusting technical parameters in stages).

[0094] Advantages of the eigenvector calculation include:

[0095] Mean and variance normalization: Standardize the matrix elements by row mean and row variance to reduce the interference of single-dimensional data anomalies (such as random expression deviations in individual bidding documents) on overall feature extraction, and improve the stability of the model in data noise scenarios. Global mean constraint: The difference operation between the difference matrix elements and the global mean can filter out the scale differences between different bidding projects (such as the natural differences in the technical complexity of different projects) and focus on the comparison of behavioral patterns of different bidders in the same project. The first eigenvector, based on exponential weighted summation, focuses on capturing typical patterns that appear frequently in the bidding behavior sequence (such as repeatedly repeated bidding strategies or technical parameter expressions), and is suitable for identifying feature clustering of habitual bid-rigging behavior. The second eigenvector, through the logarithmic geometric mean, extracts the product relationship between features, which can explore the implicit synergy between bidding behavior and winning technology (such as different bidders "coincidentally" adopting complementary technical routes in their technical solutions), and assist in discovering clues of hidden bid-rigging.

[0096] The EEMD transformation refers to integrated empirical mode decomposition.

[0097] The method of using the bidding mapping model and the first bidding behavior pattern sequence library to construct a second bidding behavior pattern sequence library and a behavior technology matching model includes:

[0098] S221, using the mapping relationship of the bidding mapping model, mapping the description statement of each bidding behavior in the first bidding behavior pattern sequence to the technical description statement of the winning bid; after the mapping operation is completed for all the description statements of the bidding behavior in the first bidding behavior pattern sequence, determining the mapped first bidding behavior pattern sequence as the corresponding second bidding behavior pattern sequence;

[0099] S222, integrating all second bidding behavior pattern sequences to obtain a second bidding behavior pattern sequence library;

[0100] S223, performing statistical analysis on the number of simultaneous occurrences of behavior types and technical numbers in the historical bid winning information set to obtain corresponding relationships between behavior types and technical numbers; and constructing a behavior-technology matching model using all corresponding relationships.

[0101] The statistical analysis of the number of simultaneous occurrences of the behavior type and the technical number in the historical bid winning information set to obtain the corresponding relationship between the behavior type and the technical number includes:

[0102] Count the number of occurrences of the combination of behavior type and technical number in each historical bid winning information set to obtain a statistical value for each combination; and preset a statistical threshold;

[0103] For each combination, determine in turn whether its statistical value is greater than the statistical threshold. If it is greater than the statistical threshold, determine that the behavior type and technical number in the combination have a corresponding relationship; if it is less than the statistical threshold, determine that the behavior type and technical number in the combination have no corresponding relationship.

[0104] The method of constructing a behavior pattern sequence pair library by using the second bidding behavior pattern sequence library includes:

[0105] S231: For each second bidding behavior pattern sequence in the second bidding behavior pattern sequence library, construct a technology number pair using the technology numbers of the adjacent winning bid solution technology description sentences contained therein;

[0106] S232: Utilize all technical number pairs constructed by the second bidding behavior pattern sequences to construct a behavior pattern sequence pair library.

[0107] The behavior types include price priority, technological advancement priority, technological reliability priority, and dynamic iterative upgrades.

[0108] The extracting of the technical number pair from the bidding behavior pattern sequence in the second bidding behavior pattern sequence may be: the second bidding behavior pattern sequence library includes the following bidding behavior pattern sequences:

[0109] Bidding behavior pattern sequence 3: Technical description statement 1 of the winning bid (technical number 4) → Technical description statement 2 of the winning bid (technical number 5);

[0110] Bidding behavior pattern sequence 6: Technical description statement 1 of the winning bid (technical number 4) → Technical description statement 2 of the winning bid (technical number 5) → Technical description statement 3 of the winning bid (technical number 6);

[0111] Bidding behavior pattern sequence 8: Technical description statement 3 of the winning bid (technical number 1) → Technical description statement 4 of the winning bid (technical number 2) → Technical description statement 5 of the winning bid (technical number 3);

[0112] Then the technical number pairs that can be extracted are: technical number 1→technical number 2, technical number 2→technical number 3, technical number 4→technical number 5, technical number 5→technical number 6.

[0113] The bidding types include: open bidding, required bidding, competitive negotiation, and inquiry and comparison.

[0114] The bidding technical requirement number is the quantitative number value of the main technical indicator requirements in the bidding proposal;

[0115] The technical number corresponding to the technical description of the winning proposal is the quantitative number value of the main technical indicators in the winning proposal;

[0116] The quantization number value can be obtained through a general text quantization encoding algorithm.

[0117] The method of analyzing and processing the bidding document information set by using the rule set to obtain a related information set includes:

[0118] S31, grouping the bidding document information using the bidding IP address to obtain a plurality of bidding document information sets; a bidding document information set includes a plurality of bidding document information having the same bidding IP address;

[0119] S32, sorting all the bidding document information in each bidding document information set according to their bidding time to obtain a corresponding bidding document information sequence;

[0120] S33, for each bidding document information sequence, performing technical correlation determination on every two adjacent bidding document information in the bidding document information sequence to obtain technical correlation relationship information of the bidding document information sequence;

[0121] S34, performing feature association recognition on the technical association relationship information of all bidding document information sequences to obtain an association information set.

[0122] The present invention enables correlation analysis between the technical route numbers in bid documents and the technical route numbers in historical bid winning information. By matching the technical descriptions of the winning bid proposals in the historical bid winning information set with the bid technical route numbers in the bid document information set, a more accurate basis for evaluating technical proposals can be provided to tendering parties. Tendering parties can refer to the strengths and weaknesses of the technical proposals in historical winning bids and, in combination with the technical route of the current bid, more scientifically assess the feasibility and innovation of the bid technical proposal, thereby improving the quality of the tendered project.

[0123] The step of performing technical correlation determination on every two adjacent bidding document information in the bidding document information sequence to obtain technical correlation relationship information of the bidding document information sequence includes:

[0124] S331, for each two adjacent bidding document information in the bidding document information sequence, obtain the two corresponding behavior types; and determine the two technology numbers corresponding to the two behavior types using a behavior technology matching model;

[0125] S332, performing a joint difference calculation on the two determined technology numbers and the corresponding bidding technology route numbers of two adjacent bidding document information to obtain a technology number difference value;

[0126] S333, determining whether the technical number difference is less than a preset first discrimination threshold; if so, determining that the two adjacent bidding document information have a technical association relationship;

[0127] The expression for the joint difference calculation is:

[0128]

[0129] Among them, bh is the technical number difference value, α1 and α2 are the two determined technical numbers, and β1 and β2 are the bidding technical route numbers of the corresponding two adjacent bidding document information.

[0130] The technical association relationship information of all bidding document information sequences is subjected to feature association identification to obtain an association information set, including:

[0131] S341, for each bidding document information having a technical association relationship, obtain the two corresponding behavior types; and determine the two technology numbers corresponding to the two behavior types using a behavior-technology matching model;

[0132] S342, constructing a to-be-matched technology number pair using the two determined technology numbers;

[0133] S343: Searching the behavior pattern sequence pair library to see whether it contains the same behavior pattern sequence pair as the to-be-matched technical number pair; if so, determining that a bidding association relationship exists between the two bidding documents corresponding to the to-be-matched technical number pair; if not, determining that no bidding association relationship exists between the two bidding documents corresponding to the to-be-matched technical number pair;

[0134] S344: Utilize all bidding document information having bidding association relationships to construct an association information set.

[0135] According to a second aspect of an embodiment of the present invention, a data association analysis device in the field of bidding is disclosed, the device comprising:

[0136] a memory storing executable program code;

[0137] a processor coupled to the memory;

[0138] The processor calls the executable program code stored in the memory to execute the data association analysis method in the bidding field.

[0139] According to a third aspect of an embodiment of the present invention, a computer-storable medium is disclosed, wherein the computer-storable medium stores computer instructions. When the computer instructions are called by a computer, the computer instructions are used to execute the data association analysis method in the bidding field.

[0140] According to a fourth aspect of an embodiment of the present invention, an information data processing terminal is disclosed, which is used to implement the data association analysis method in the bidding field.

[0141] The foregoing is merely an embodiment of the present invention and is not intended to limit the present invention. It will be apparent to those skilled in the art that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are intended to be included within the scope of the claims of the present invention.

Claims

1. A data association analysis method in the field of bidding, characterized by: include: S1, collecting and obtaining a set of historical bidding information; the set of historical bidding information includes a set of historical bidding information and a set of historical winning bid information; the set of historical bidding information includes historical bidding information; the historical bidding information includes a description of bidding behavior; the set of historical winning bid information includes historical winning bid information; the historical winning bid information includes a winning bid type, behavior type, a technical description of the winning bid solution, and a corresponding technical number; S2, performing rule mining processing on the bidding history information set to obtain a rule set; the rule set includes a behavior technology matching model and a behavior pattern sequence pair library; S3, collecting and obtaining a bidding document information set; the bidding document information set includes bidding document information; the bidding document information includes bidding time, bidding IP address, behavior type, bidding technical route number and bidding type; S4: Process the bidding document information set using the rule set to obtain a related information set.

2. The data association analysis method in the bidding field according to claim 1, characterized in that: The rule mining process is performed on the bidding history information set to obtain a rule set, including: S21, constructing a first bidding behavior pattern sequence library and a bidding mapping model using the bidding history information set; S22, using the bidding mapping model and the first bidding behavior pattern sequence library to construct a second bidding behavior pattern sequence library and a behavior technology matching model; S23, using the second bidding behavior pattern sequence library, construct a behavior pattern sequence pair library; the rule set includes a behavior technology matching model and a behavior pattern sequence pair library.

3. The data association analysis method in the bidding field according to claim 2, characterized in that: The method of constructing a first bidding behavior pattern sequence library and a bidding mapping model by utilizing the historical bidding information set and the historical winning bid information set includes: S211, performing an extraction operation on each piece of historical bidding information in the historical bidding information set to obtain a description statement of the corresponding bidding behavior; S212: sorting the bidding behavior description sentences extracted from each historical bidding information according to the appearance time of the description sentences to obtain a first bidding behavior pattern sequence of the historical bidding information; the first bidding behavior pattern sequence is a directed sequence of bidding behavior description sentences; S213, integrating the first bidding behavior pattern sequences of all historical bidding information to obtain a first bidding behavior pattern sequence library; S214, performing semantic similarity calculation on the description of the bidding behavior in the first bidding behavior pattern sequence and the technical description of the winning bid solution in the historical winning bid information set to obtain a technical similarity value; S215: Determine whether the technical similarity value is greater than a set similarity threshold; if so, establish a mapping relationship between the technical description of the winning bid in the historical winning bid information set and the bidding behavior corresponding to the description of the bidding behavior in the first bidding behavior pattern sequence; S216: Utilize all established mapping relationships to construct a bidding mapping model.

4. The data association analysis method in the bidding field according to claim 2, characterized in that: The semantic similarity calculation of the description statement of the bidding behavior of the first bidding behavior pattern sequence and the technical description statement of the winning bid solution in the historical winning bid information set to obtain a technical similarity value includes: A first description matrix is ​​constructed using the description statements of all bidding behaviors in the first bidding behavior pattern sequence; a row vector of the first description matrix is ​​a vector of a description statement of a bidding behavior in the first bidding behavior pattern sequence; A second description matrix is ​​constructed using all technical description statements of the winning bid solutions in the historical winning bid information set; a row vector of the second description matrix is ​​a vector of a technical description statement of a winning bid solution in the historical winning bid information set; Difference calculation is performed on the first description matrix and the second description matrix to obtain a technical similarity value.

5. The data association analysis method in the bidding field according to claim 4, characterized in that: The performing difference calculation on the first description matrix and the second description matrix to obtain a technical similarity value includes: Subtracting the first description matrix from the second description matrix to obtain a difference matrix; Performing eigenvector calculation on the difference matrix to obtain a first eigenvector and a second eigenvector; Performing EEMD transformation on the second eigenvector to obtain a eigentransformation vector; The first feature vector and the feature transformation vector are fused and calculated to obtain a technical similarity value.

6. The data association analysis method in the bidding field according to claim 5, characterized in that: The expression for calculating the eigenvector is: Among them, p1 i and p2 i are the i-th elements of the first and second eigenvectors, respectively, P ij is the element of the i-th row and j-th column of the difference matrix, α i is the mean of the i-th row of the difference matrix, β i is the variance of the i-th row of the difference matrix, λ is the mean of all elements of the difference matrix, and N is the column dimension of the difference matrix; The expression of the fusion calculation is: Among them, yf is the technical similarity value, M is the total number of elements of the first eigenvector, H 2i is the i-th element of the feature transformation vector.

7. The data association analysis method in the bidding field according to claim 1, characterized in that: The method of analyzing and processing the bidding document information set by using the rule set to obtain a related information set includes: S31, grouping the bidding document information using the bidding IP address to obtain a plurality of bidding document information sets; a bidding document information set includes a plurality of bidding document information having the same bidding IP address; S32, sorting all the bidding document information in each bidding document information set according to their bidding time to obtain a corresponding bidding document information sequence; S33, for each bidding document information sequence, performing technical correlation determination on every two adjacent bidding document information in the bidding document information sequence to obtain technical correlation relationship information of the bidding document information sequence; S34, performing feature association recognition on the technical association relationship information of all bidding document information sequences to obtain an association information set.

8. A data association analysis device in the field of bidding, characterized in that: The device comprises: a memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the data association analysis method in the bidding field according to any one of claims 1 to 7.

9. A computer storable medium, characterized in that The computer storable medium stores computer instructions, and when the computer instructions are called by a computer, they are used to execute the data association analysis method in the bidding field according to any one of claims 1 to 7.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the data association analysis method in the bidding field according to any one of claims 1 to 7.

Images