CN-122022968-A - Intelligent early warning method and system for abnormal bidding behaviors

CN122022968ACN 122022968 ACN122022968 ACN 122022968ACN-122022968-A

Abstract

The invention discloses an intelligent early warning method and system for abnormal bidding behaviors, and relates to the technical field of data processing and analysis. The method comprises the steps of firstly obtaining full flow data of bidding projects, generating a suspicious spot list through preliminary anomaly detection, then using the bidding as a query starting point, mining cross-project association information, constructing a dynamic event map with target projects as cores, revealing cross-project association risk information, and finally carrying out fusion research and judgment on the suspicious spot list and the cross-project risk information to generate early warning prompt information. The method solves the problems that the prior art is difficult to identify cross-project and cross-space deep collusion relations and has insufficient scene adaptability.

Inventors

WANG RUI
PU JINGJING
HUANG JINSONG

Assignees

四川建设网有限责任公司

Dates

Publication Date: 20260512
Application Date: 20260415

Claims (10)

1. An intelligent early warning method for abnormal bidding behaviors is characterized by comprising the following steps: Acquiring full-flow data of a target item, wherein the full-flow data comprises bidder information, a bidding file, quotation data and bidder behavior logs; Performing preliminary abnormality detection based on the full-flow data, and generating a suspicious point list according to a detection result; Mining association information from an association data source based on a preset multidimensional association rule by taking a bidder of the target item as a query starting point, wherein the association data source comprises historical participation items of the bidder; Constructing a dynamic event map centering on the target item according to the association information, and revealing inter-item association risk information based on the dynamic event map; And carrying out fusion research and judgment on the suspicious point list and the cross-project associated risk information to generate early warning prompt information.
2. The intelligent early warning method for abnormal bidding behavior according to claim 1, wherein the preliminary abnormal detection is performed based on the full-flow data, and the generating of the suspicious point list according to the detection result comprises: detecting whether the bidding behavior characteristics of different bidders are the same or similar, and recording the detected abnormal behavior and the bidders, the abnormal types and the evidence abstract related to the abnormal behavior to a suspicious point list, wherein the abnormal behavior comprises at least one of the following steps: a preset numerical rule exists among the bidding offers; The text similarity of bidding documents of different bidders in preset key chapters reaches a set threshold; There is a coincidence in the network addresses and/or hardware device identifications used by different bidders to submit bid files.
3. The intelligent pre-warning method for abnormal bidding behavior according to claim 1, wherein mining the association information from the association data source based on a preset multidimensional association rule by using the bidding for the target item as a query starting point comprises: Extracting the artificial business information of each bidder as a keyword, and executing inquiry in the associated data source to acquire the associated entity and the associated type of the bidder to form a static associated candidate set, wherein the associated type comprises a stock right control relationship and a personnel job-holding relationship; Extracting bidding behavior characteristics of each bidder in the target item, and using the bidding behavior characteristics as a query condition, and querying similar bidding subjects and similar behaviors thereof with the same or similar bidding behavior characteristics in historical participation items of the bidders to form a behavior association candidate set, wherein the similar behaviors comprise equipment sharing behaviors, quotation similarity and bidding text similarity; And merging and de-duplicating the static association candidate set and the behavior association candidate set to construct an association information set containing the bidder, the association entity and association type thereof, the similar bidding subject and similar behavior and source evidence thereof.
4. The intelligent pre-warning method for abnormal bidding behavior according to claim 3, wherein the constructing a dynamic event map centered on the target item according to the association information comprises: Establishing a static subgraph reflecting inherent association among the subjects by taking bidders, association entities and similar bidding subjects in the association information set as graph nodes and the association type as graph edges; Taking the target item and the historical participation item as item nodes, taking the bidder and the similar bidding subject as bidder nodes, taking bidding behaviors as bidding relation edges connecting the item nodes and the bidder nodes, taking the similar behaviors as behavior relation edges connecting the bidder nodes, marking common bidding times, bidding time distribution and quotation rules as edge attributes on the behavior relation edges, and constructing an event subgraph centering on the target item; And performing association mapping on the bidder nodes of the event subgraph and the graph nodes of the static subgraph, establishing cross-layer association edges, and generating a dynamic event map taking the target item as a center.
5. The intelligent pre-warning method for abnormal bidding behavior according to claim 4, wherein the revealing cross-project associated risk information based on the dynamic event profile comprises: Performing graph traversal and topology analysis on the dynamic event graph to extract a risk feature subgraph; In the risk characteristic subgraph, bidders which meet the condition that direct or indirect association paths exist between bidder nodes and have common bidding behaviors in at least two or more historical participation items are divided into the same risk group: Distributing associated risk values for each risk partner based on the number of associated paths, the number of common bids, the characteristic similarity of bidding behavior and the size of the partner, and packaging the identified risk partner and associated evidence and associated risk values into cross-item associated risk information; the risk feature subgraphs comprise dense connection subgraphs and indirect association path subgraphs, wherein the dense connection subgraphs are high-density connection structures formed by common use equipment, common association personnel or a plurality of historical common bidding relations among bidder nodes, and the indirect association path subgraphs are connection structures in which no direct edges exist between two bidder nodes but accessible paths are formed by one or more intermediate nodes.
6. The method for intelligently warning abnormal bidding behaviors according to claim 1 or 5, wherein the fusing and judging the suspicious point list and the cross-item associated risk information to generate warning prompt information comprises the following steps: combining the bidders in the doubtful point list with the bidders related to the cross-item associated risk information to obtain abnormal bidders; and generating structured early warning prompt information containing a risk main body and evidence according to the abnormal behavior of the abnormal bidder in the target item and the cross-item associated risk information.
7. The intelligent pre-warning method for abnormal bidding behavior according to claim 1, wherein the full-process data further comprises review process data, the method further comprising: extracting review abnormal characteristics from the review process data; performing collaborative analysis on the suspicious point list, the cross-project associated risk information and the evaluation abnormal characteristics, and outputting comprehensive early warning information; The evaluation process data comprises at least one of expert scoring details, evaluation committee discussion records, audio and video monitoring texts and expert browsing taggant behavior logs.
8. The method of claim 7, wherein the extracting the review anomaly from the review process data of the target item comprises at least one of: The scoring of each expert is standardized, outlier experts with scores which are significantly deviated from the scoring mean value of the bidders are identified, and the outlier degree of the outlier experts is calculated; natural language processing is carried out on the bid evaluation audio and video text, and a speaking segment for dialect, excessive question or implicit bid intention is extracted; and analyzing the time distribution of browsing each bidding document by an expert, and identifying abnormal browsing behaviors.
9. An intelligent early warning system for bidding transaction flow, comprising: The data acquisition module is used for acquiring the whole flow data of the target item, wherein the whole flow data comprises bidder information, a bidding file, quotation data and a bidding behavior log; the suspicious point preliminary screening module is used for carrying out preliminary abnormality detection based on the whole flow data and generating a suspicious point list according to the detection result; The association mining module is used for mining association information from an association data source based on a preset multidimensional association rule by taking a bidder of the target item as a query starting point, wherein the association data source comprises historical participation items of the bidder; the risk analysis module is used for constructing a dynamic event map centering on a target item according to the association information and revealing cross-item association risk information based on the dynamic event map; And the fusion judging module is used for carrying out fusion judgment on the suspicious point list and the cross-project associated risk information and generating early warning prompt information.
10. The intelligent early warning system of claim 9, further comprising a visual display module for displaying the dynamic event map, the early warning prompt information, and the comprehensive early warning information in a visual form.

Description

Intelligent early warning method and system for abnormal bidding behaviors Technical Field The invention relates to the technical field of data processing and analysis, in particular to an intelligent early warning method and system for abnormal bidding behaviors. Background Along with the popularization of the electronic bidding platform, links such as bidding document compiling, submitting, bid opening, bid evaluation and the like are gradually realized on-line, so that the transaction efficiency and standardization level are greatly improved. However, at the same time, the electronic device also enables abnormal behaviors such as collusion bidding, bidder ring and the like to be more hidden, and brings new challenges for bidding supervision work. The traditional bidding abnormal behavior detection method usually focuses on explicit feature recognition in a single project, for example, bidding offers of different bidders show numerical rules such as arithmetic series, fixed proportion floating down, and the like, technical bidding texts have too high similarity, bidding documents are submitted by using the same hardware/network identification, and the like. The method has a certain effect when detecting string mark behaviors in the same project by adopting simple technical means, but is difficult to realize effective early warning when facing specialized, cross-regional and cross-project collusion and partner. The existing detection method only analyzes the current project or the contemporaneous project, and can not sense the historical behaviors of bidders in different projects, different regions and different periods. Bidder ring group partner often adopts 'tour tactics' -alternately accompany mark companies and cross combinations in different projects, and under the single project view, the behaviors are not obviously abnormal and cannot be effectively identified. Disclosure of Invention The invention provides an intelligent early warning method and system for bidding abnormal behaviors, which are used for solving the problems that the existing bidding abnormal detection technology is insufficient in hidden association mining capability and is difficult to identify cross-project and cross-space-time deep collusion relations. The invention is realized by the following technical scheme: The invention provides an intelligent early warning method for abnormal bidding behaviors, which comprises the following steps: Acquiring full-flow data of a target item, wherein the full-flow data comprises bidder information, a bidding file, quotation data and bidder behavior logs; Performing preliminary abnormality detection based on the full-flow data, and generating a suspicious point list according to a detection result; Mining association information from an association data source based on a preset multidimensional association rule by taking a bidder of the target item as a query starting point, wherein the association data source comprises historical participation items of the bidder; Constructing a dynamic event map centering on the target item according to the association information, and revealing inter-item association risk information based on the dynamic event map; And carrying out fusion research and judgment on the suspicious point list and the cross-project associated risk information to generate early warning prompt information. Further, the performing preliminary anomaly detection based on the full-flow data, generating a suspicious point list according to a detection result, includes: detecting whether the bidding behavior characteristics of different bidders are the same or similar, and recording the detected abnormal behavior and the bidders, the abnormal types and the evidence abstract related to the abnormal behavior to a suspicious point list, wherein the abnormal behavior comprises at least one of the following steps: a preset numerical rule exists among the bidding offers; The text similarity of bidding documents of different bidders in preset key chapters reaches a set threshold; There is a coincidence in the network addresses and/or hardware device identifications used by different bidders to submit bid files. Further, the mining the association information from the association data source based on the preset multidimensional association rule by using the bidder of the target item as a query starting point comprises the following steps: Extracting the artificial business information of each bidder as a keyword, and executing inquiry in the associated data source to acquire the associated entity and the associated type of the bidder to form a static associated candidate set, wherein the associated type comprises a stock right control relationship and a personnel job-holding relationship; Extracting bidding behavior characteristics of each bidder in the target item, and using the bidding behavior characteristics as a query condition, and querying similar bidding subjects and similar behaviors t