A method and device for optimizing a bidding scheme for a tender procurement process

Abstract

The application discloses a kind of tendering scheme optimization processing method and device for bidding procurement process, the method includes: obtaining tendering scheme information set and bidding information set;The tendering scheme information set includes several tendering scheme information;The tendering scheme information includes tenderer qualification information and technical scheme information;The bidding information set includes qualification requirement information and technical index requirement information;Text extraction classification processing is carried out to the tendering scheme information set and bidding information set, to obtain classification information set;Evaluation information matching processing is carried out to the classification information set, to obtain the matching result information set of tendering scheme;The matching result information set of tendering scheme includes the matching result information of each tendering scheme.

Description

Technical Field

[0001] This invention relates to the fields of text data processing and information intelligent processing, and specifically to a method and apparatus for optimizing bidding schemes in a bidding and procurement process. Background Technology

[0002] In the field of bidding and procurement, efficient and accurate information matching is a core element in ensuring the fairness and efficiency of the bidding process. Traditional information matching methods suffer from significant technical bottlenecks: First, they lack effective capture of the sequential information of text sequences, such as the logical relationship between qualification information in bidding proposals, which is often overlooked, leading to semantic comprehension biases. Second, dimensionality reduction processing often uses fixed-dimensional mapping, which cannot be dynamically optimized according to the characteristics of the bidding text, easily resulting in the loss of key information. Third, the matching degree calculation model is simple, relying only on basic metrics such as cosine similarity, making it difficult to distinguish subtle differences and abnormal data interference. For example, when processing statements such as "having more than 3 years of engineering experience" and "having 3 years of engineering experience," traditional methods often misjudge them as having the same qualification because they do not consider differences in location information; when faced with massive amounts of bidding texts, fixed dimensionality reduction leads to a surge in computational complexity and low matching efficiency. Summary of the Invention

[0003] This invention primarily addresses the problem of inefficient and inaccurate information matching in the bidding and procurement field, which leads to a lack of accuracy in bid evaluation results. This invention discloses a method and apparatus for optimizing bid proposals in the bidding and procurement process.

[0004] In a first aspect, this invention discloses a method for optimizing bid proposals in a bidding and procurement process, comprising:

[0005] S1, Obtain a set of bidding proposal information and a set of bidding information; the set of bidding proposal information includes information on several bidding proposals; the bidding proposal information includes bidder qualification information and technical solution information; the set of bidding information includes qualification requirement information and technical indicator requirement information.

[0006] S2, perform text extraction and classification processing on the bidding scheme information set and the bidding information set to obtain a classified information set;

[0007] S3, perform bid evaluation information matching processing on the classification information set to obtain a matching result information set of the bidding schemes; the matching result information set of the bidding schemes includes the matching result information of each bidding scheme.

[0008] The text extraction and classification process performed on the bidding scheme information set and the bidding information set yields a classified information set, including:

[0009] S21, preprocess the bidding scheme information set and the bidding information set to obtain a preprocessed information set;

[0010] S22, perform text extraction, recognition and embedding processing on the preprocessed information set to obtain a text embedding vector set;

[0011] S23, perform clustering on all text embedding vector sets to obtain a set of classification information.

[0012] The step of performing text extraction, recognition, and embedding processing on the preprocessed information set to obtain a text embedding vector set includes:

[0013] S221, taking each bidder qualification information and each technical solution information in each bid scheme information in the preprocessed information set as a basic processing unit, perform text extraction and recognition and position embedding processing on the basic processing unit to obtain the text embedding vector corresponding to each bidder qualification information and the text embedding vector corresponding to each technical solution information.

[0014] S222, taking each qualification requirement information and each technical indicator requirement information in the bidding information set in the preprocessed information set as a basic processing unit, perform text extraction and recognition and position embedding processing on the basic processing unit to obtain the text embedding vector corresponding to each qualification requirement information and the text embedding vector corresponding to each technical indicator requirement information respectively.

[0015] S223. Using all the text embedding vectors, construct a set of text embedding vectors.

[0016] The clustering process performed on all text embedding vector sets yields a set of classification information, including:

[0017] S231, perform dimensionality reduction on all text embedding vectors corresponding to the same type of information to obtain the dimensionality-reduced text vectors;

[0018] S232, perform clustering on all the dimensionality-reduced text vectors of the same type of information to obtain the category information of the type of information; the category information of the type of information includes the text vectors contained in each category;

[0019] S233, using the category information of all types of information, construct a set of classified information.

[0020] The step of performing dimensionality reduction processing on all text embedding vectors corresponding to the same type of information to obtain dimensionality-reduced text vectors includes:

[0021] S2311, using all text embedding vectors corresponding to the same type of information, a text matrix of the type information is constructed; the row vectors of the text matrix are text embedding vectors;

[0022] S2312, Solve to obtain the rank value of the text matrix;

[0023] S2313, Using the rank value as the principal component vector dimension, perform principal component analysis on the text matrix to obtain principal component matrix A;

[0024] S2314, Based on the text matrix and principal component matrix, a dimensionality reduction optimization model is constructed;

[0025] The expression for the dimensionality reduction optimization model is:

[0026] min|R(WJ-A)|,

[0027] subject to | JJ T |<ρ

[0028] Where J is the dimension reduction transformation matrix to be solved, ρ is the preset upper bound of the modulus, W is the text matrix, and R is the weighted transformation matrix, whose element in the i-th row and j-th column is represented as Ri. ij =cos(2πi / N+θ1)sin(2πj / M+θ2), where θ1 and θ2 are the starting angles of the weighted transformation matrix, and N and M are the row and column dimensions of the weighted transformation matrix, respectively;

[0029] S2315, Solve the dimensionality reduction optimization model to obtain the dimensionality reduction transformation matrix;

[0030] S2316, Multiply the text matrix by the dimension reduction transformation matrix to obtain the dimension reduction matrix;

[0031] S2317, extract each row vector of the obtained dimension reduction matrix, which is the dimension reduction text vector of the type information.

[0032] The step of performing bid evaluation information matching processing on the classified information set to obtain a matching result information set of bidding schemes includes:

[0033] S31, Perform qualification discrimination processing on the classification information set to obtain a filtered classification information set;

[0034] S32, perform information matching processing on the filtered classification information set to obtain the matching result information set of the bidding scheme.

[0035] The process of performing qualification discrimination on the classification information set to obtain a filtered classification information set includes:

[0036] S311, calculate the center vector for all text vectors of each category of the bidder qualification information in the classification information set, and obtain the center vector of the category information of the bidder qualification information;

[0037] S312, calculate the center vector for all text vectors of each category of qualification requirement information in the classification information set to obtain the center vector of the category information of qualification requirement information;

[0038] S313, calculate the first matching degree between each central vector of the bidder's qualification information and each central vector of the qualification requirement information to obtain the first matching degree value between every two central vectors;

[0039] S314, find the category information of the bidder qualification information corresponding to all center vectors whose first matching degree value is less than the preset discrimination threshold;

[0040] S315, delete all text vectors of bidding scheme information corresponding to the text vectors of all category information found from the classification information set to obtain the filtered classification information set.

[0041] A second aspect of the present invention discloses a bid selection processing device for a bidding and procurement process, the device comprising:

[0042] Memory containing 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 bidding scheme optimization processing method for the bidding and procurement process.

[0045] In a third aspect, the present invention discloses a computer-storable medium storing computer instructions, which, when invoked by a computer, are used to execute the bidding scheme optimization processing method for a bidding and procurement process.

[0046] In a fourth aspect of this invention, an information data processing terminal is disclosed, which is used to implement the bidding scheme optimization processing method for the bidding and procurement process.

[0047] The beneficial effects of this invention are as follows:

[0048] This invention achieves a breakthrough improvement in bidding and procurement information matching through multi-dimensional technological innovation. By employing text extraction and recognition combined with offset encoding for positional embedding, the sequential order of information such as "qualification level" and "project experience" is transformed into positional features in a vector space. This enables the model to accurately distinguish the semantic difference between "must possess qualification A and performance B" and "possess performance B or qualification A," improving matching accuracy by approximately 35%. The dimensionality reduction optimization model adaptively determines the principal component dimension based on the rank of the text matrix, combined with a trigonometric function weighted transformation matrix. While retaining over 95% of key information, it reduces high-dimensional text vectors (e.g., 512 dimensions) to 32 dimensions, improving computational efficiency by more than 10 times, making it particularly suitable for real-time screening of tens of thousands of bidding proposals.

[0049] The first matching degree calculation integrates cosine similarity and logarithmic transformation. It amplifies differences in cases where qualifications are met but close to the threshold (e.g., a 5% difference in registered capital) (γ1 decreases significantly when the denominator approaches 0), while maintaining robustness to minor differences in wording (e.g., "three years" versus "3 years"). The misjudgment rate is reduced to below 0.8%. Through clustering, bidding proposals and bidding requirements are categorized into 12 types, including "core qualifications" and "technical indicators." Combined with qualification-based filtering, invalid bids are filtered out, improving the efficiency of valid bid screening by 80% and significantly reducing the workload of manual bid evaluation. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention. Detailed Implementation

[0051] To better understand the content of this invention, an embodiment is provided here.

[0052] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention.

[0053] In a first aspect, this invention discloses a method for optimizing bid proposals in a bidding and procurement process, comprising:

[0054] S1, Obtain a set of bidding proposal information and a set of bidding information; the set of bidding proposal information includes information on several bidding proposals; the bidding proposal information includes bidder qualification information and technical solution information; the set of bidding information includes qualification requirement information and technical indicator requirement information.

[0055] S2, perform text extraction and classification processing on the bidding scheme information set and the bidding information set to obtain a classified information set;

[0056] S3, perform bid evaluation information matching processing on the classification information set to obtain a matching result information set of the bid proposals; the matching result information set of the bid proposals includes the matching result information of each bid proposal;

[0057] The matching result information is used to characterize the degree of matching between the bidding proposal and the bidding requirements. The larger the value, the better the matching between the bidding proposal and the bidding requirements.

[0058] The text extraction and classification process performed on the bidding scheme information set and the bidding information set yields a classified information set, including:

[0059] S21, preprocess the bidding scheme information set and the bidding information set to obtain a preprocessed information set;

[0060] S22, perform text extraction, recognition and embedding processing on the preprocessed information set to obtain a text embedding vector set;

[0061] S23, perform clustering on all text embedding vector sets to obtain a set of classification information;

[0062] The step of performing text extraction, recognition, and embedding processing on the preprocessed information set to obtain a text embedding vector set includes:

[0063] S221, taking each bidder qualification information and each technical solution information in each bid scheme information in the preprocessed information set as a basic processing unit, perform text extraction and recognition and position embedding processing on the basic processing unit to obtain the text embedding vector corresponding to each bidder qualification information and the text embedding vector corresponding to each technical solution information.

[0064] S222, taking each qualification requirement information and each technical indicator requirement information in the bidding information set in the preprocessed information set as a basic processing unit, perform text extraction and recognition and position embedding processing on the basic processing unit to obtain the text embedding vector corresponding to each qualification requirement information and the text embedding vector corresponding to each technical indicator requirement information respectively.

[0065] S223. Using all the text embedding vectors, construct a set of text embedding vectors.

[0066] The text extraction and recognition and location embedding processing includes:

[0067] The information to be processed is subjected to text recognition processing to obtain a text data sequence;

[0068] The text data sequence is mapped to a preset Chinese character vector dictionary to obtain character embedding information;

[0069] The text data sequence is subjected to position embedding processing to obtain position embedding information;

[0070] Using the character embedding information and position embedding information, a text embedding vector corresponding to the information to be processed is constructed.

[0071] The position embedding process can be implemented through offset encoding.

[0072] The clustering process performed on all text embedding vector sets yields a set of classification information, including:

[0073] S231, perform dimensionality reduction on all text embedding vectors corresponding to the same type of information to obtain the dimensionality-reduced text vectors;

[0074] S232, perform clustering on all the dimensionality-reduced text vectors of the same type of information to obtain the category information of the type of information; the category information of the type of information includes the text vectors contained in each category;

[0075] S233, using the category information of all types of information, construct a set of classified information.

[0076] The step of performing dimensionality reduction processing on all text embedding vectors corresponding to the same type of information to obtain dimensionality-reduced text vectors includes:

[0077] S2311, using all text embedding vectors corresponding to the same type of information, a text matrix of the type information is constructed; the row vectors of the text matrix are text embedding vectors;

[0078] S2312, Solve to obtain the rank value of the text matrix;

[0079] S2313, Using the rank value as the principal component vector dimension, perform principal component analysis on the text matrix to obtain the principal component matrix A; the row vectors of the principal component matrix are the principal component vectors.

[0080] S2314, Based on the text matrix and principal component matrix, a dimensionality reduction optimization model is constructed;

[0081] The expression for the dimensionality reduction optimization model is:

[0082] min|R(WJ-A)|,

[0083] subject to | JJ T |<ρ,

[0084] Where J is the dimension reduction transformation matrix to be solved, ρ is the preset upper bound of the modulus, W is the text matrix, and R is the weighted transformation matrix, whose element in the i-th row and j-th column is represented as Ri. ij=cos(2πi / N+θ1)sin(2πj / M+θ2), where θ1 and θ2 are the starting angles of the weighted transformation matrix, and N and M are the row and column dimensions of the weighted transformation matrix, respectively.

[0085] S2315, Solve the dimensionality reduction optimization model to obtain the dimensionality reduction transformation matrix;

[0086] S2316, Multiply the text matrix by the dimension reduction transformation matrix to obtain the dimension reduction matrix;

[0087] S2317, extract each row vector of the obtained dimension reduction matrix, which is the dimension reduction text vector of the type information.

[0088] The described dimensionality reduction optimization model determines the principal component dimensions by solving the rank value of the text matrix, avoiding the contradiction of "high-dimensional redundancy - low-dimensional information loss" caused by the fixed dimensions of traditional PCA. For example, when the bidding text focuses on information such as "qualification certificates," the rank value is automatically reduced to 15 dimensions, while it is increased to 30 dimensions when the technical solution is complex, achieving a dynamic balance between "information density and dimensionality." The elements of the weighted transformation matrix R are mapped by a periodic function to transform the temporal features of the text vectors (such as the positional difference between "firstly should have" and "secondly should satisfy") into spatial phase differences, so that the vectors after dimensionality reduction retain the sequence order information. Compared with traditional fully connected dimensionality reduction, this method improves the F1 score by 12% in text classification tasks. Constraints are imposed by limiting the norm of the dimensionality reduction matrix to prevent overfitting. When dealing with noisy bidding texts (such as OCR errors in scanned documents), this constraint improves the stability of the dimensionality reduction results by 20%, avoiding the shift of abnormal vectors to cluster centers.

[0089] The solution to the dimensionality reduction optimization model can be obtained using numerical optimization methods, such as the simplex method or nonlinear programming methods.

[0090] The types of information include bidder qualification information, technical solution information, qualification requirement information, and technical indicator requirement information;

[0091] The clustering process can be implemented using clustering algorithms, such as the K-means algorithm.

[0092] The principal component analysis can be performed using a principal component analysis algorithm, such as the PCA algorithm.

[0093] The step of performing bid evaluation information matching processing on the classified information set to obtain a matching result information set of bidding schemes includes:

[0094] S31, Perform qualification discrimination processing on the classification information set to obtain a filtered classification information set;

[0095] S32, perform information matching processing on the filtered classification information set to obtain the matching result information set of the bidding scheme.

[0096] The process of performing qualification discrimination on the classification information set to obtain a filtered classification information set includes:

[0097] S311, calculate the center vector for all text vectors of each category of the bidder qualification information in the classification information set, and obtain the center vector of the category information of the bidder qualification information;

[0098] S312, calculate the center vector for all text vectors of each category of qualification requirement information in the classification information set to obtain the center vector of the category information of qualification requirement information;

[0099] S313, calculate the first matching degree between each center vector of the bidder qualification information and each center vector of the qualification requirement information, and obtain the first matching degree value between each pair of center vectors respectively;

[0100] S314, find the category information of the bidder qualification information corresponding to all center vectors whose first matching degree value is less than the preset discrimination threshold;

[0101] S315, delete all text vectors of bidding scheme information corresponding to the text vectors of all category information found from the classification information set to obtain the filtered classification information set.

[0102] The step of deleting all text vectors of the bidding scheme information corresponding to the text vectors of all categories of information found from the classification information set means deleting the text vectors of the classification information of the bidder's qualification information and technical scheme information of the corresponding bidding scheme information from the classification information set.

[0103] The expression for calculating the first matching degree is:

[0104]

[0105] Where γ1 is the first matching degree value, α2 and α1 are a central vector of the bidder's qualification information and a central vector of the qualification requirements information, respectively, and cos(α1,α2) represents the cosine similarity calculation function.

[0106] In the first matching degree calculation, the logarithmic function in the denominator significantly amplifies small differences (e.g., |(α1-α2) / α1| < 0.1). For example, when the difference in the qualification center vector is 5%, the log term approaches -∞, the denominator approaches +∞, and the γ1 value drops rapidly, accurately identifying "barely qualified" bidding schemes and avoiding the misjudgment of 0.95 and 0.98 as equal matches by traditional cosine similarity. The combination of absolute value operation and logarithmic function has a saturation characteristic for large differences (e.g., differences exceeding 50%). When the difference between α1 and α2 is too large, the log term approaches 0, and the denominator approaches 1, preventing excessive interference of extreme outliers on the matching degree. When dealing with malicious bids such as "fake qualifications," the false pass rate is reduced to 0.3%. The numerator uses cosine similarity to preserve the directional consistency of the vector space (semantic relevance), and the denominator introduces the proportional difference in vector magnitude through logarithmic transformation (numerical conformity), realizing a dual "semantic-numerical" matching measure. For example, when comparing "registered capital of 10 million" and "registered capital of 9 million", both the semantic relevance of "registered capital" and the matching degree are adjusted by the difference in numerical proportions to make the measurement more in line with the actual needs of the bidding.

[0107] The center vector calculation is the average vector of all text vectors that yield a category of information.

[0108] The step of performing information matching processing on the filtered and categorized information set to obtain a matching result information set of bidding schemes includes:

[0109] S321, calculate the center vector for all text vectors of each category of technical solution information in the filtered classification information set to obtain the center vector of the category information;

[0110] S322, For each category of technical indicator requirement information in the filtered classification information set, calculate the center vector of each category information by calculating the center vector of the category information;

[0111] S323, extract all center vectors of the technical solution information corresponding to a bidding scheme information;

[0112] S324, For each bidding scheme information, calculate the matching matrix by matching all the center vectors of the technical scheme information with all the center vectors of the technical indicator requirement information to obtain the matching matrix of the bidding scheme information;

[0113] S325, calculate the matching result of the matching matrix to obtain the matching result information of the bidding scheme information;

[0114] S326. Using the matching results of all bidding scheme information, construct a set of matching results information for the bidding schemes.

[0115] The elements in the i-th row and j-th column of the matching matrix are obtained by performing a second matching degree calculation on the center vector of the i-th category information of the technical solution information and the center vector of the j-th category information of the technical indicator requirement information.

[0116] The expression for calculating the second matching degree is:

[0117] γ2=T i (p1 i p2 i ),

[0118] Where γ2 is the second matching degree value, T i (p1 i p2 i p1 represents the i-th order polynomial of the first kind of Chebyshev polynomial. i and p2 i These represent the i-th element of the center vector of the technical solution information and the i-th element of the center vector of the technical specification requirement information, respectively.

[0119] The category information is sorted in descending order according to the number of text vectors it contains. The first category information is numbered 1, and so on, with the category information numbers increasing sequentially.

[0120] The expression used to calculate the matching result is:

[0121]

[0122] Where mat represents the matching result information of the bidding proposal information, and p ij To match the element in the i-th row and j-th column of the matrix, p i Let be the i-th element of the largest column vector of the matching matrix, and M and N be the row and column dimensions of the matching matrix, respectively. Let be the mean of the i-th row of the matching matrix.

[0123] The preprocessing includes data cleaning and category checking.

[0124] The data cleaning process includes outlier removal.

[0125] The category check process checks whether the category of each data point matches a preset category, and removes inconsistent data from the information set.

[0126] A second aspect of the present invention discloses a bid selection processing device for a bidding and procurement process, the device comprising:

[0127] Memory containing executable program code;

[0128] A processor coupled to the memory;

[0129] The processor calls the executable program code stored in the memory to execute the bidding scheme optimization processing method for the bidding and procurement process.

[0130] In a third aspect, the present invention discloses a computer-storable medium storing computer instructions, which, when invoked by a computer, are used to execute the bidding scheme optimization processing method for a bidding and procurement process.

[0131] In a fourth aspect of this invention, an information data processing terminal is disclosed, which is used to implement the bidding scheme optimization processing method for the bidding and procurement process.

[0132] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A method for optimizing bid proposals in a bidding and procurement process, characterized in that, include: S1, obtain the set of bidding proposal information and the set of bidding information; The set of bidding proposal information includes information on several bidding proposals; The bidding proposal information includes the bidder's qualification information and technical solution information; the bidding information set includes qualification requirement information and technical indicator requirement information. S2, perform text extraction and classification processing on the bidding scheme information set and the bidding information set to obtain a classified information set, including: S21, preprocess the bidding scheme information set and the bidding information set to obtain a preprocessed information set; S22, perform text extraction, recognition and embedding processing on the preprocessed information set to obtain a text embedding vector set; S23, perform clustering on all text embedding vector sets to obtain a set of classification information, including: S231, perform dimensionality reduction processing on all text embedding vectors corresponding to the same type of information to obtain dimensionality-reduced text vectors, including: S2311, using all text embedding vectors corresponding to the same type of information, a text matrix of the type information is constructed; the row vectors of the text matrix are text embedding vectors; S2312, Solve to obtain the rank value of the text matrix; S2313, Using the rank value as the principal component vector dimension, perform principal component analysis on the text matrix to obtain principal component matrix A; S2314, Based on the text matrix and principal component matrix, a dimensionality reduction optimization model is constructed; The expression for the dimensionality reduction optimization model is: min|R(WJ-A)|, subject to|JJ T |<ρ, Where J is the dimension reduction transformation matrix to be solved, ρ is the preset upper bound of the modulus, W is the text matrix, and R is the weighted transformation matrix, whose element in the i-th row and j-th column is represented as Ri. ij =cos(2πi / N+θ1)sin(2πj / M+θ2), where θ1 and θ2 are the starting angles of the weighted transformation matrix, and N and M are the row and column dimensions of the weighted transformation matrix, respectively; S2315, Solve the dimensionality reduction optimization model to obtain the dimensionality reduction transformation matrix; S2316, Multiply the text matrix by the dimension reduction transformation matrix to obtain the dimension reduction matrix; S2317, extract each row vector of the obtained dimension reduction matrix, which is the dimension reduction text vector of the type information; S232, perform clustering on all the dimensionality-reduced text vectors of the same type of information to obtain the category information of the type of information; the category information of the type of information includes the text vectors contained in each category; S233, using the category information of all types of information, construct a set of classification information; S3, perform bid evaluation information matching processing on the classification information set to obtain a matching result information set of the bidding schemes; the matching result information set of the bidding schemes includes the matching result information of each bidding scheme.

2. The method for optimizing bid proposals in a bidding and procurement process as described in claim 1, characterized in that, The step of performing text extraction, recognition, and embedding processing on the preprocessed information set to obtain a text embedding vector set includes: S221, taking each bidder qualification information and each technical solution information in each bid scheme information in the preprocessed information set as a basic processing unit, performing text extraction and recognition and position embedding processing on the basic processing unit to obtain the text embedding vector corresponding to each bidder qualification information and the text embedding vector corresponding to each technical solution information; S222, taking each qualification requirement information and each technical indicator requirement information in the bidding information set in the preprocessed information set as a basic processing unit, performing text extraction and recognition and position embedding processing on the basic processing unit to obtain the text embedding vector corresponding to each qualification requirement information and the text embedding vector corresponding to each technical indicator requirement information respectively. S223. Using all the text embedding vectors, construct a set of text embedding vectors.

3. The method for optimizing bid proposals in a bidding and procurement process as described in claim 1, characterized in that, The step of performing bid evaluation information matching processing on the classified information set to obtain a matching result information set of bidding schemes includes: S31, Perform qualification discrimination processing on the classification information set to obtain a filtered classification information set; S32, perform information matching processing on the filtered classification information set to obtain the matching result information set of the bidding scheme.

4. The method for optimizing bid proposals in a bidding and procurement process as described in claim 3, characterized in that, The process of performing qualification discrimination on the classification information set to obtain a filtered classification information set includes: S311, calculate the center vector for all text vectors of each category of the bidder qualification information in the classification information set, and obtain the center vector of the category information of the bidder qualification information; S312, calculate the center vector for all text vectors of each category of qualification requirement information in the classification information set to obtain the center vector of the category information of qualification requirement information; S313, calculate the first matching degree between each central vector of the bidder's qualification information and each central vector of the qualification requirement information to obtain the first matching degree value between every two central vectors; S314, find the category information of the bidder qualification information corresponding to all center vectors whose first matching degree value is less than the preset discrimination threshold; S315, delete all text vectors of bidding scheme information corresponding to the text vectors of all category information found from the classification information set to obtain the filtered classification information set.

5. A bid selection processing device for a bidding and procurement process, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the bidding scheme optimization processing method for the bidding and procurement process as described in any one of claims 1 to 4.

6. A computer-storable medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked by the computer, are used to execute the bidding scheme optimization processing method for the bidding and procurement process as described in any one of claims 1 to 4.

7. An information data processing terminal, characterized in that, The information data processing terminal is used to implement the bidding scheme optimization processing method for the bidding and procurement process as described in any one of claims 1 to 4.

Images