Intelligent identification system and method for enclosing and stringing labels based on artificial intelligence

Abstract

The invention discloses a system and a method for intelligently identifying a bid invitation on the basis of artificial intelligence, relates to the technical field of intelligent bid invitation identification, and solves the technical problem that the execution time of a bid invitation project is easily delayed as a bid invitation behavior is not identified in time in a bid invitation process in the prior art. The method comprises the following steps: analyzing a price floating rate between a quoted price of each associated enterprise in an associated enterprise group and a bid invitation quoted price; respectively counting the number of associated enterprises whose price floating rates are higher than or lower than a floating threshold value; extracting a plurality of bid winning information from the historical data, and analyzing the historical bid winning rate of the associated enterprise group; analyzing the similarity between the bidding files finally participating in bidding, and analyzing the similarity ratio of the bidding files; the number of the associated enterprises with the price floating rate higher than or lower than the floating threshold value, the historical bid winning rate of the associated enterprise group and the similar proportion of the bidding files are synthesized, so that bid enclosing or bid confusion behaviors in the bidding process can be timely and effectively found, and a fair and impartial bidding and tendering competition environment is promoted.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent identification systems for bid rigging and collusion, and specifically relates to an intelligent identification system and method for bid rigging and collusion based on artificial intelligence. Background Art

[0002] The identification and prevention of bid rigging and collusion in the bidding process has always been a hot topic of research in the bidding field at home and abroad.

[0003] The invention patent application with publication number CN112561670A discloses an intelligent identification system for bid rigging and bid collusion, and the intelligent identification system includes a project bidding platform, a data processing subsystem, a bid rigging identification subsystem and a bid collusion identification subsystem. The intelligent identification system of the present invention includes a project bidding platform, a data processing subsystem, a bid rigging identification subsystem and a bid collusion identification subsystem. It integrates the identification functions of online bidding and bid rigging and bid collusion, wherein the data processing subsystem is used to obtain, identify and associate the bidders' information; and under the bid rigging identification subsystem, the bidders' joint bidding behavior, price joint behavior and bid evaluation price control behavior are identified.

[0004] The above-mentioned prior art proposes to identify bid rigging and collusion in bidding activities after the enterprise wins the bid, rather than promptly identifying bid rigging and collusion during the bidding process, which can easily delay the execution time of the bidding project; therefore, the present invention proposes an intelligent identification system and method for bid rigging and collusion based on artificial intelligence. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an intelligent identification system and method for bid rigging based on artificial intelligence, which is used to solve the technical problem that the prior art fails to identify bid rigging in a timely manner during the bidding process, which easily delays the execution time of the bidding project.

[0006] To achieve the above-mentioned purpose, the first aspect of the present invention provides an artificial intelligence-based intelligent identification system for bid rigging, comprising an intelligent identification module for bid rigging, and a bidding information analysis module and a bidding information management platform connected thereto;

[0007] Bidding information analysis module: used to extract basic information of bidding enterprises, analyze the correlation index between bidding enterprises, and obtain several groups of related enterprises; analyze the price fluctuation rate between the quotations of each related enterprise in the related enterprise group and the bidding quotation; and count the number of related enterprises whose price fluctuation rate is higher or lower than the floating threshold; wherein, the bidding information is extracted from the bidding information management platform;

[0008] Extract some winning bid information from historical data and analyze the historical winning bid rates of the associated enterprise groups;

[0009] Also, analyze the similarity between the bidding documents of the affiliated enterprises that ultimately participate in the bidding, and analyze the similarity ratio of the bidding documents;

[0010] Intelligent identification module for bid rigging and collusion: used to determine whether there is bid rigging and collusion in the bidding activity; if yes, the bidding activity will be stopped immediately; if no, the bidding activity will continue.

[0011] Preferably, the bidding information management platform stores bidding information and winning bid information, and encrypts the bidding information; wherein each bidding enterprise holds a hardware encryption device issued by a certification body to manage its unique digital certificate; the bidding document is encrypted by the public key of the digital certificate and decrypted by the corresponding private key when the bid is opened; the bidding information includes the basic information of the bidding enterprise and the quotation of the bidding enterprise.

[0012] The bidding information management platform of the present invention manages all bidding / tendering information in a unified manner, including the release of bidding information, the submission of bidding documents and the announcement of winning bids. It also ensures the uniqueness and non-replicability of digital certificates by equipping each bidding enterprise with a hardware encryption device issued by an authoritative certification body. The entire process is open and transparent while ensuring privacy, thereby enhancing the trust of all parties involved.

[0013] Preferably, the extracting of basic information of the bidding enterprises and analyzing the correlation indexes between the bidding enterprises to obtain a number of related enterprise groups include:

[0014] The basic information of the bidding companies is divided according to information attributes to obtain several attribute information groups, and the correlation coefficients between the same attribute information groups of each bidding company are analyzed and marked as Xi; the correlation index P between any two bidding companies is calculated by the formula P=Σ(αi×Xi); if the correlation index is greater than the correlation threshold, the company is a related company, and several related company groups are obtained; wherein, i represents the number of the attribute information group, i=0, 1,…, N, N is a positive integer; αi is the weight coefficient.

[0015] Preferably, the analysis of the correlation coefficients between the same attribute information groups of the bidding enterprises includes:

[0016] Determine whether the attribute elements of the same dimension in the same attribute information group are the same. If yes, the attribute element is marked as 1; if not, the attribute element is marked as 0; then obtain the binary vector of the attribute information group, and calculate the correlation coefficient Xi between the same attribute information groups by the formula Xi=Ai / Bi; where Ai is the number of elements in the binary vector that are 1, and Bi is the total number of elements in the binary vector.

[0017] The present invention can accurately capture the similarities and potential associations between enterprises by breaking down the basic information of bidding enterprises into multiple attribute information groups and comparing the attribute elements of the same dimension in these groups one by one. This method is more scientific and objective than the traditional judgment based on fuzzy impressions or single indicators, and effectively avoids misjudgment and omission. It also uses binary vectors to represent attribute information groups and calculates association coefficients through simple mathematical operations, which greatly simplifies the data processing process and improves calculation efficiency. By identifying related enterprises, it helps to reveal potential market monopolies and unfair competition behaviors, thereby maintaining the fairness and healthy competition environment of the bidding market.

[0018] Preferably, the analyzing the price fluctuation rate between the quotations of each associated enterprise in the associated enterprise group and the bidding quotations includes:

[0019] The price fluctuation rate JFL is calculated by the formula JFL=(TBJ-ZBJ) / ZBJ; wherein TBJ is the quotation of the bidding enterprise; ZBJ is the quotation of the tendering enterprise.

[0020] Preferably, extracting a number of winning bid information from historical data and analyzing the historical winning bid rates of the associated enterprise group includes:

[0021] Filter out the associated enterprises from the winning bid information, count the number of associated enterprises, mark them as SC, and calculate the historical winning bid rate YB of the associated enterprise group through the formula YB=SC / WZ; wherein WZ is the number of other enterprises in the winning bid information except the associated enterprise group.

[0022] The present invention obtains the historical bid winning rate by counting the historical bid winning times of the associated enterprises and performing ratio processing with the number of bid winnings of other enterprises except the enterprises in the associated enterprise group; the calculation of the historical bid winning rate reveals the actual performance of the associated enterprises in the bidding activities, which helps to identify possible market monopoly or unfair competition behaviors, such as the potential rotation of bid winning.

[0023] Preferably, the analyzing the similarity between the bidding documents of the affiliated enterprises that eventually participate in the bidding includes:

[0024] Step S1: Divide the bidding document into chapters and extract keywords / sentences from the core chapters; wherein the core chapters are chapters with substantive content;

[0025] Step S2: taking any bidding document as a comparison document, performing semantic similarity analysis on the keywords / sentences of the same core sections between the comparison document and other bidding documents; performing weighted summation on the semantic similarities of the core sections to obtain the similarity between the comparison document and the bidding document;

[0026] Step S3: Determine whether the similarity is greater than the similarity threshold; if yes, classify the bid document and the comparison document into one category; if no, determine whether the number of remaining unclassified bid documents exceeds one; if yes, reselect any bid document from the remaining unclassified bid documents as the comparison document and return to step S2; if no, end the similarity comparison.

[0027] The present invention divides the bidding documents into chapters and focuses on the core chapters with substantive content, thereby ensuring that the analysis is focused on the most relevant information, which not only improves efficiency but also ensures the accuracy of the analysis; extracting keywords / sentences from the core chapters helps to quickly locate key information points. This method can help reviewers quickly understand the key content of the bidding documents and also helps automated tools to perform subsequent similarity analysis.

[0028] Preferably, the analysis of similarity ratios of bidding documents includes:

[0029] The number of classified bidding documents is counted and compared with the total number of bidding documents finally participating in the bidding to obtain the similarity ratio of the bidding documents.

[0030] Preferably, the determining whether there is bid-rigging or collusion in the bidding activity includes:

[0031] The number of affiliated enterprises whose price fluctuation rate is higher or lower than the fluctuation threshold, the historical winning rate of the affiliated enterprise group and the similarity ratio of the bidding documents are input into the bid rigging identification model, and the bid rigging situation is output; among them, the bid rigging identification model is constructed based on the BP neural network model.

[0032] Preferably, an intelligent identification method of an artificial intelligence-based bid-rigging intelligent identification system comprises the following steps:

[0033] Step 1: Extract the basic information of the bidding enterprises, analyze the correlation index between the bidding enterprises, and obtain several related enterprise groups;

[0034] Step 2: Extract the quotations of the associated enterprise group, analyze the price fluctuation rate between the quotations of each associated enterprise in the associated enterprise group and the bidding quotations; and count the number of associated enterprises whose price fluctuation rate is higher or lower than the floating threshold;

[0035] Step 3: Extract some winning bid information from historical data and analyze the historical winning bid rates of the associated enterprise groups;

[0036] Step 4: Analyze the similarity between the final bidding documents and the similarity ratio of the bidding documents;

[0037] Step 5: Identify bid rigging in bidding activities based on the number of affiliated companies whose price fluctuation rates are above or below the fluctuation threshold, the historical winning rate of the affiliated company group and the similarity ratio of the bid documents.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1. The present invention extracts basic information of bidding enterprises, analyzes the correlation index between bidding enterprises, obtains several groups of related enterprises, reveals possible correlation relationships between bidding enterprises, and analyzes the correlation information between related enterprises, such as the number of related enterprises with price fluctuation rates higher or lower than the floating threshold, the historical winning rate of the related enterprise group and the similarity ratio of the bidding documents. Through the obtained correlation information, potential bid-rigging behaviors can be timely and effectively identified, and measures can be taken in time to ensure the fairness of the bidding activities;

[0040] 2. The present invention extracts quotations from a group of associated enterprises, and analyzes the price fluctuation rate between the quotations of each associated enterprise in the group and the bidding quotations; respectively counts the number of associated enterprises whose price fluctuation rate is higher or lower than the floating threshold; extracts a number of winning bid information from historical data, and analyzes the historical winning bid rate of the associated enterprise group; analyzes the similarity between the bidding documents that ultimately participate in the bidding, and analyzes the similarity ratio of the bidding documents; identifies bid rigging and collusion in bidding activities through three-aspect analysis; and helps to automatically identify and crack down on bid rigging or collusion in the bidding process, thereby promoting a fairer competition environment and protecting the rights and interests of legitimate bidders. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 It is a schematic diagram of the system structure of the present invention;

[0043] Figure 2 It is a schematic diagram of the process of the present invention;

[0044] Figure 3 It is a schematic diagram of the similarity analysis process of the bidding documents of the present invention. DETAILED DESCRIPTION

[0045] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] See also Figure 1 , the first aspect of the present invention provides an artificial intelligence-based intelligent identification system for bid rigging and collusion, including an intelligent identification module for bid rigging and collusion, and a bidding information analysis module and a bidding information management platform connected thereto;

[0047] The bidding information management platform stores bidding information and winning bid information, and encrypts the bidding information; each bidding enterprise holds a hardware encryption device issued by a certification body to manage its unique digital certificate; the bidding documents are encrypted with the public key of the digital certificate and decrypted with the corresponding private key when the bid is opened; the bidding information includes the basic information of the bidding enterprise and the quotation of the bidding enterprise.

[0048] The bidding information analysis module extracts the basic information of the bidding enterprises of a bidding project from the bidding information management platform, analyzes the correlation index between the bidding enterprises, and obtains several related enterprise groups;

[0049] Specifically, the basic information of the bidding companies is divided according to information attributes to obtain several attribute information groups, and it is determined whether the attribute elements of the same dimension in the same attribute information group are the same. If yes, the attribute element is marked as 1; if not, the attribute element is marked as 0; then a binary vector of the attribute information group is obtained, and the correlation coefficient Xi between the same attribute information groups is calculated by the formula Xi=Ai / Bi; wherein Ai is the number of elements in the binary vector that are 1, and Bi is the total number of elements in the binary vector; the attribute element is the keyword of the attribute information; i represents the number of the attribute information group, i=0, 1,…, N, and N is a positive integer.

[0050] It should be noted that the basic information of the bidding company includes shareholder information, ultimate beneficiary information, management personnel registration information, core personnel information, controlled enterprise information, foreign investment information, financing information, industrial and commercial change information, bidding / tendering information and supplier information.

[0051] The correlation index P between any two bidding enterprises is calculated by the formula P=Σ(αi×Xi); if the correlation index is greater than the correlation threshold, they are related enterprises, and several related enterprise groups are obtained; αi is the weight coefficient.

[0052] The present invention evaluates the degree of correlation between two enterprises through the correlation index; the degree of correlation is mainly closely related to the correlation coefficient between each same attribute information group and the weight coefficient of each attribute information group; if the correlation coefficients between the attribute information groups are large, it indicates that the degree of correlation between the two enterprises is high, that is, the two enterprises are associated and the correlation index is large.

[0053] For example: In this embodiment, it is assumed that the basic information of three bidding companies A, B, and C is extracted, and the correlation between the bidding companies is analyzed by extracting two groups of attribute information groups, namely, shareholder information and supplier information;

[0054] Extract keywords from the above attribute information group of each enterprise;

[0055] For example: Enterprise A: shareholder information: [Zhang San, Li Si], supplier information: [D Company, E Company]; Enterprise B: shareholder information: [Zhang San, Zhao Liu], supplier information: [D Company, E Company]; Enterprise C: shareholder information: [Wang Yi, Li Wu], supplier information: [H Company, E Company]; For each attribute group, the keywords between two enterprises are compared to obtain the binary vector of the attribute information group; among which, the shareholder information group is marked as 0, and the supplier information group is marked as 1; assuming α0 = 0.5, α1 = 0.5; the correlation index threshold is 0.6;

[0056] The binary vector of the shareholder information group of Enterprise A and Enterprise B is: [1, 0], the correlation coefficient X0 = 0.5; the binary vector of the supplier information is: [1, 1], the correlation coefficient X1 = 1; the correlation index between Enterprise A and Enterprise B is P = 0.5 × 0.5 + 0.5 × 1 = 0.75, which is greater than the correlation index threshold, then Enterprise A is associated with Enterprise B;

[0057] The binary vector of the shareholder information group of Enterprise A and Enterprise C is: [0, 0], the correlation coefficient X0 = 0; the binary vector of the supplier information is: [0, 1], the correlation coefficient X1 = 0.5; the correlation index between Enterprise A and Enterprise C is P = 0.5 × 0.5 = 0.25, which is less than the correlation index threshold, so Enterprise A and Enterprise C are not related;

[0058] The binary vector of the shareholder information group of Enterprise B and Enterprise C is: [0, 0], and the correlation coefficient X0=0; the binary vector of the supplier information is: [0, 1], and the correlation coefficient X1=0.5; the correlation index between Enterprise B and Enterprise C is P=0.5×0.5=0.25, which is less than the correlation index threshold, so Enterprise B and Enterprise C are not associated.

[0059] Extract the quotations of the associated enterprise group and calculate the price fluctuation rate JFL through the formula JFL=(TBJ-ZBJ) / ZBJ; TBJ is the quotation of the bidding enterprise; ZBJ is the quotation of the tendering enterprise; count the number of associated enterprises whose price fluctuation rate is higher or lower than the floating threshold respectively.

[0060] The present invention evaluates the fluctuation of the bid price of a bidding enterprise relative to the bid price of a tendering enterprise through the price fluctuation rate; the price fluctuation is mainly closely related to the bid price and the tender price. If the bid price is larger and the tender price is smaller, the price fluctuation rate is larger, indicating that the bid price is larger than the tender price and the bid price is much higher than the tender price.

[0061] Extract information of several historical bid-winning enterprises from historical data and analyze the historical bid-winning rates of related enterprise groups;

[0062] Specifically, several winning bid information are extracted from historical data, related enterprises are screened out from the winning bid information, the number of related enterprises is counted and marked as SC, and the historical winning bid rate YB of the related enterprise group is calculated by the formula YB=SC / WZ; wherein WZ is the number of other enterprises in the winning bid information except the related enterprise group.

[0063] The present invention analyzes the historical bid-winning probability of the associated enterprise group through the historical bid-winning rate; the historical bid-winning probability of the associated enterprise group is mainly closely related to the number of associated enterprises in the bid-winning information and the number of other enterprises in the bid-winning information except the associated enterprise group. If the number of associated enterprises in the bid-winning information is large and the number of other enterprises in the bid-winning information except the associated enterprise group is small, it indicates that the bid-winning probability of the associated enterprise group is high, that is, the historical bid-winning rate is large, and there may be signs of taking turns to win the bid.

[0064] Also, analyze the similarity between the bidding documents of the affiliated enterprises that ultimately participate in the bidding, and analyze the similarity ratio of the bidding documents;

[0065] See also Figure 2 Specifically, step S1: divide the bidding document into chapters and extract keywords / sentences from the core chapters; wherein the core chapters are chapters with substantive content;

[0066] Step S2: taking any bidding document as a comparison document, performing semantic similarity analysis on the keywords / sentences of the same core sections between the comparison document and other bidding documents; performing weighted summation on the semantic similarities of the core sections to obtain the similarity between the comparison document and the bidding document;

[0067] Step S3: Determine whether the similarity is greater than the similarity threshold; if yes, classify the bid document and the comparison document into one category; if no, determine whether the number of remaining unclassified bid documents exceeds one; if yes, reselect any bid document from the remaining unclassified bid documents as the comparison document and return to step S2; if no, end the similarity comparison.

[0068] For example, the three bidding documents of the three affiliated enterprises A, B, and C are Q, K, and V respectively; if the Q bidding document is used as the comparison document, similarity analysis is performed with the two bidding documents K and V respectively; assuming that only the similarity of the K bidding document is greater than the similarity threshold, then K and Q are similar bidding documents and are classified into one category; the remaining V document is not analyzed;

[0069] If the Q bid document is used as a comparison document, similarity analysis is performed with the K and V bid documents respectively; assuming that the similarities of the K and V bid documents are not greater than the similarity threshold, they are not classified; and the remaining V document is not analyzed; therefore, the bid documents of the associated companies A, B, and C are not similar.

[0070] The number of classified bidding documents is counted and compared with the total number of bidding documents that finally participate in the bidding to obtain the similarity ratio of the bidding documents; among which, the bidding documents that finally participate in the bidding are the bidding documents submitted by the enterprise for the last time before the bidding deadline.

[0071] It should be noted that the bidding documents should be divided into: each bidding document should be divided into chapters, such as "Project Overview", "Technical Plan", "Implementation Plan", etc.; core chapters should be extracted: determine which chapters are core chapters, that is, the parts containing substantive content, such as "Technical Plan" and "Implementation Plan"; keywords / sentences should be extracted from each core chapter, including but not limited to noun phrases, verb phrases and important sentences.

[0072] Intelligent identification module for bid rigging and collusion: used to determine whether there is bid rigging and collusion in the bidding activity; if yes, the bidding activity will be stopped immediately; if no, the bidding activity will continue.

[0073] The number of affiliated enterprises whose price fluctuation rate is higher or lower than the fluctuation threshold, the historical winning rate of the affiliated enterprise group and the similarity ratio of the bidding documents are input into the bid rigging identification model, and the bid rigging situation is output; among them, the bid rigging identification model is constructed based on the BP neural network model.

[0074] For example: extract the number of affiliated enterprises with price fluctuation rates higher or lower than the floating threshold, the historical winning rate of the affiliated enterprise group and the similarity ratio of the bidding documents, and the corresponding bid-rigging and bid-rigging situations from the historical bid-rigging and bid-rigging analysis data; the number of affiliated enterprises with price fluctuation rates higher or lower than the floating threshold, the historical winning rate of the affiliated enterprise group and the similarity ratio of the bidding documents are used as the input data of the BP neural network model, and the bid-rigging and bid-rigging situations are used as the output data of the BP neural network model;

[0075] The input data and output data are divided into two parts, namely the training data set and the verification data set. The BP neural network model is trained by the training data set, and the verification data set is used to verify the training results of the BP neural network model, and finally a high-precision bid rigging identification model is obtained.

[0076] See also Figure 3 The second aspect of the present invention proposes an intelligent identification method of an artificial intelligence-based bid-rigging intelligent identification system, comprising the following steps:

[0077] Step 1: Extract the basic information of the bidding enterprises, analyze the correlation index between the bidding enterprises, and obtain several related enterprise groups;

[0078] Step 2: Extract the quotations of the associated enterprise group, analyze the price fluctuation rate between the quotations of each associated enterprise in the associated enterprise group and the bidding quotations; and count the number of associated enterprises whose price fluctuation rate is higher or lower than the floating threshold;

[0079] Step 3: Extract some winning bid information from historical data and analyze the historical winning bid rates of the associated enterprise groups;

[0080] Step 4: Analyze the similarity between the final bidding documents and the similarity ratio of the bidding documents;

[0081] Step 5: Identify bid rigging in bidding activities based on the number of affiliated companies whose price fluctuation rates are above or below the fluctuation threshold, the historical winning rate of the affiliated company group and the similarity ratio of the bid documents.

[0082] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0083] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An artificial intelligence-based intelligent identification system for bid rigging, characterized in that: It includes an intelligent identification module for bid rigging and collusion, as well as a bidding information analysis module and a bidding information management platform connected thereto; Bidding information analysis module: used to extract basic information of bidding enterprises, analyze the correlation index between bidding enterprises, obtain several related enterprise groups; analyze the price fluctuation rate between the quotations of each related enterprise in the related enterprise group and the bidding quotations; The number of affiliated enterprises whose price fluctuation rate is higher or lower than the floating threshold is counted respectively; among which, the bidding information is extracted from the bidding information management platform; Extract some winning bid information from historical data and analyze the historical winning bid rates of the associated enterprise groups; Also, analyze the similarity between the bidding documents of the affiliated enterprises that ultimately participate in the bidding, and analyze the similarity ratio of the bidding documents; Intelligent identification module for bid rigging and collusion: used to determine whether there is bid rigging and collusion in the bidding activity; if yes, the bidding activity will be stopped immediately; if no, the bidding activity will continue.

2. According to claim 1, the intelligent identification system for bid rigging based on artificial intelligence is characterized in that: The bidding information management platform stores bidding information and winning bid information, and encrypts the bidding information; each bidding enterprise holds a hardware encryption device issued by a certification body to manage its unique digital certificate; the bidding documents are encrypted with the public key of the digital certificate and decrypted with the corresponding private key when the bid is opened; the bidding information includes the basic information of the bidding enterprise and the quotation of the bidding enterprise.

3. The artificial intelligence-based intelligent identification system for bid rigging according to claim 2 is characterized in that: The basic information of the bidding enterprises is extracted, and the correlation index between the bidding enterprises is analyzed to obtain several groups of related enterprises, including: The basic information of the bidding companies is divided according to information attributes to obtain several attribute information groups, and the correlation coefficients between the same attribute information groups of each bidding company are analyzed and marked as Xi; the correlation index P between any two bidding companies is calculated by the formula P=Σ(αi×Xi); if the correlation index is greater than the correlation threshold, the company is a related company, and several related company groups are obtained; wherein, i represents the number of the attribute information group, i=0, 1,…, N, N is a positive integer; αi is the weight coefficient.

4. The artificial intelligence-based intelligent identification system for bid rigging according to claim 3 is characterized in that: The analysis of the correlation coefficients between the same attribute information groups of the bidding enterprises includes: Determine whether the attribute elements of the same dimension in the same attribute information group are the same. If yes, the attribute element is marked as 1; if not, the attribute element is marked as 0; then obtain the binary vector of the attribute information group, and calculate the correlation coefficient Xi between the same attribute information groups by the formula Xi=Ai / Bi; where Ai is the number of elements in the binary vector that are 1, and Bi is the total number of elements in the binary vector.

5. The artificial intelligence-based intelligent identification system for bid rigging according to claim 3 is characterized in that: The analysis of the price fluctuation rate between the quotations of each associated enterprise in the associated enterprise group and the bidding quotations includes: The price fluctuation rate JFL is calculated by the formula JFL=(TBJ-ZBJ) / ZBJ; wherein TBJ is the quotation of the bidding enterprise; ZBJ is the quotation of the tendering enterprise.

6. The artificial intelligence-based intelligent identification system for bid rigging according to claim 1 is characterized in that: The step of extracting a number of winning bid information from historical data and analyzing the historical winning bid rates of the associated enterprise group includes: Filter out the associated enterprises from the winning bid information, count the number of associated enterprises, mark them as SC, and calculate the historical winning bid rate YB of the associated enterprise group through the formula YB=SC / WZ; wherein WZ is the number of other enterprises in the winning bid information except the associated enterprise group.

7. The artificial intelligence-based intelligent identification system for bid rigging according to claim 1 is characterized in that: The analysis of the similarity between the bidding documents of the affiliated enterprises that eventually participate in the bidding includes: Step S1: Divide the bidding document into chapters and extract keywords / sentences from the core chapters; wherein the core chapters are chapters with substantive content; Step S2: taking any bidding document as a comparison document, performing semantic similarity analysis on the keywords / sentences of the same core sections between the comparison document and other bidding documents; performing weighted summation on the semantic similarities of the core sections to obtain the similarity between the comparison document and the bidding document; Step S3: Determine whether the similarity is greater than the similarity threshold; if yes, classify the bid document and the comparison document into one category; if no, determine whether the number of remaining unclassified bid documents exceeds one; if yes, reselect any bid document from the remaining unclassified bid documents as the comparison document and return to step S2; if no, end the similarity comparison.

8. The artificial intelligence-based intelligent identification system for bid rigging according to claim 7 is characterized in that: The similarity ratio of the bidding documents analyzed includes: The number of classified bidding documents is counted and compared with the total number of bidding documents finally participating in the bidding to obtain the similarity ratio of the bidding documents.

9. The artificial intelligence-based intelligent identification system for bid rigging according to claim 1 is characterized in that: The determination of whether there is bid-rigging or collusion in the bidding activities includes: The number of affiliated enterprises whose price fluctuation rate is higher or lower than the fluctuation threshold, the historical winning rate of the affiliated enterprise group and the similarity ratio of the bidding documents are input into the bid rigging identification model, and the bid rigging situation is output; among them, the bid rigging identification model is constructed based on the BP neural network model.

10. An intelligent identification method of an artificial intelligence-based intelligent identification system for bid rigging, characterized in that: The following steps are involved: Step 1: Extract the basic information of the bidding enterprises, analyze the correlation index between the bidding enterprises, and obtain several related enterprise groups; Step 2: Extract the quotations of the associated enterprise group and analyze the price fluctuation rate between the quotations of each associated enterprise in the associated enterprise group and the bidding quotations; Count the number of affiliated companies whose price fluctuation rate is higher or lower than the fluctuation threshold respectively; Step 3: Extract some winning bid information from historical data and analyze the historical winning bid rates of the associated enterprise groups; Step 4: Analyze the similarity between the final bidding documents and the similarity ratio of the bidding documents; Step 5: Identify bid rigging in bidding activities based on the number of affiliated companies whose price fluctuation rates are above or below the floating threshold, the historical winning rate of the affiliated company group and the similarity ratio of the bid documents.

Images