An information management method and system based on an electronic bidding transaction platform

Abstract

The application discloses an information management method and system based on an electronic bidding transaction platform, and relates to the technical field of information management. By constructing a multi-dimensional and multi-level information management system, the potential correlation between bidding units and the behavior mode are deeply mined and intelligently identified, thereby effectively improving the ability of the electronic bidding platform in risk prevention and control and abnormal behavior identification. By fusing multi-source heterogeneous data such as business registration information, bidding history, project contract and the like, and introducing graph modeling and semantic analysis technology, the application can accurately identify key risk factors such as legal person penetration, historical cooperation and highly similar bidding documents, and then quantitatively evaluate the risk correlation strength between bidding units, and form a visual graph and an auxiliary auditing mechanism, thereby improving the judgment efficiency of supervisors.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information management, and in particular to an information management method and system based on an electronic bidding transaction platform. BACKGROUND

[0002] With the continuous advancement of "Internet+" and digital government affairs, the electronic bidding platform, as an important information tool for government procurement, engineering construction and large project management, has played an important role in improving bidding efficiency, reducing transaction costs, and enhancing transparency. Especially in the fields of construction engineering, infrastructure construction, and government procurement, the electronic bidding platform has gradually replaced the traditional offline operation mode, realizing online processing of project announcement, qualification examination, file uploading, bid opening, and contract filing. At the same time, with the continuous development of new generation information technologies such as blockchain, big data, and artificial intelligence, how to carry out deep information mining and risk control based on platform data has become one of the key directions for the evolution and upgrading of the current electronic bidding platform.

[0003] Although the existing electronic bidding platform has basic capabilities of data collection and online transaction, there are still significant deficiencies in the compliance review and risk warning of bidding activities. On the one hand, most current systems only stay at the level of static collection and basic comparison of bidding information, lacking dynamic modeling and structured identification of deep information such as historical association, equity relationship, and cooperation behavior between bidding units, making it difficult to timely detect implicit risk behaviors such as fake bidding, bid rigging, and bid stringing.

[0004] On the other hand, the existing technology lacks semantic layer analysis and behavior pattern recognition of bidding document content, and cannot accurately identify the coordination behavior between suspected bidders from the perspectives of text similarity, term overlap, and structure template. In addition, the lack of efficient visual expression methods also makes it difficult for supervisors to quickly obtain complex association paths and potential risks, increasing the difficulty and subjectivity of manual review. SUMMARY

[0005] In view of the problems of weak association information recognition and difficult risk behavior detection in existing electronic bidding technology, the present application provides an information management method and system based on an electronic bidding transaction platform.

[0006] Therefore, the problem to be solved by the present application is the identification of deep relationship between bidding units and intelligent early warning of behavior risks.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present application provides an information management method based on an electronic bidding transaction platform, which comprises the following steps: S1, collecting data information of bidding units from the electronic bidding platform, and performing standardization and structuring processing on the data information to establish a basic data information node; S2, based on the structured data information, using rule reasoning and entity matching technology to construct a graph structure between the bidding units, including explicit or implicit correlation; S3, using natural language processing technology to perform semantic feature extraction and similarity comparison on the bidding documents of the current project, and identifying suspicious behavior patterns of multiple units; S4, fusing and modeling the structural correlation information and the semantic behavior analysis results, calculating the risk correlation strength between the bidding units, and generating a risk level label; S5, presenting the risk correlation strength and the risk level label in a graph visualization manner, supporting the supervisor to view the correlation path and behavior trajectory of the bidding units, and providing an artificial review interface for auxiliary judgment, forming a closed-loop processing mechanism.

[0009] As a preferred scheme of the information management method based on the electronic bidding transaction platform, the specific implementation process of step S1 comprises:

[0010] The original information data set of the bidding units on the electronic bidding platform is collected, and the original information data set comprises business registration information, historical bidding records, project bidding information and contract clause texts.

[0011] A field template is preset, and the unstructured fields in the original information data set are reorganized and regularized, so that the legal person information, shareholder structure, historical project keywords and contract clause abstracts are reconstructed into a structured field set, and a field confidence parameter is labeled.

[0012] The natural language content in the structured field set is subjected to a semantic unification operation, and the contract clause texts, the project bidding information and the service clauses in the contract clause texts are subjected to standardization processing based on semantic normalization rules to obtain a uniform semantic expression field set.

[0013] Based on the structured field set and the uniform semantic expression field set, a bidding unit basic node is established, and the bidding unit basic node comprises an identification field, a field confidence parameter and a semantic expression field, and serves as basic metadata of a unit node in a knowledge graph.

[0014] As a preferred scheme of the information management method based on the electronic bidding transaction platform, the specific implementation process of step S2 comprises:

[0015] Extract the legal person name field, contact information field and registered address field in the basic information node to form a candidate entity set, and take the entity with a field confidence greater than a preset entity confidence threshold as a high-confidence entity set;

[0016] According to the legal name field and the registered capital field in the high-confidence entity set, under the condition of a preset equity reasoning rule, the control path relationship strength between the two bidding units is calculated, and if the control path relationship strength between the two bidding units is greater than or equal to a preset control path relationship strength threshold, the two bidding unit basic nodes are connected as equity control relationship edges;

[0017] The historical bidding project keyword field and the bid-winning project time field in the high-confidence entity set are compared jointly, and according to the project participation time overlap rate and the keyword semantic similarity, it is judged whether the two bidding units have a historical joint bidding behavior, and a joint behavior relationship edge is constructed;

[0018] The graph structure composed of the unit node and the equity control relationship edge and the joint behavior relationship edge is taken as the initial knowledge graph between the bidding units, and the node inherits the identification field and the semantic expression field in the graph structure.

[0019] As a preferred scheme of the information management method based on the electronic bidding transaction platform, the specific implementation process of step S3 further includes:

[0020] Text segmentation processing is performed on each file in the current project bidding file set, a semantic substructure set composed of bidding clause response segments, technical parameter segments and enterprise introduction segments is extracted, and each semantic segment is labeled with a paragraph position number and an original unit;

[0021] Based on the semantic substructure set, a context embedding model is used to generate a vector representation of each semantic segment, and a semantic comparison matrix is constructed by comparing the semantic segment vectors submitted by different bidding units in pairs according to a set window length.

[0022] All elements in the semantic comparison matrix are evaluated, and if there are more than three similar segments between any pair of units with a similarity higher than a preset semantic consistency threshold, the pair of units is marked as a semantic highly consistent set.

[0023] According to the distribution density of similar paragraphs and the template structure proportion, a semantic repetition degree index is calculated for the semantic structure matching of all unit pairs in the semantic highly consistent set.

[0024] As a preferred scheme of the information management method based on the electronic bidding transaction platform, the specific implementation process of step S4 includes:

[0025] Based on the path set between any two bidding unit basic node corresponding units in the initial knowledge graph between the bidding units, the path length, path type distribution and associated node density are extracted to form a structure association vector;

[0026] The semantic repetition index of any two bidding unit basic node corresponding units is retrieved to construct a semantic behavior vector, including semantic repetition density and template structure proportion;

[0027] A fusion model is constructed to calculate the risk association strength value of any two bidding unit basic node corresponding units, the input of the fusion model is the structure association vector and the semantic behavior vector, the output is the risk strength score between 0 and 1, and the risk strength threshold value is set, and the risk level label is generated according to the score;

[0028] All risk strength scores are summarized to form an inter-unit risk matrix.

[0029] As a preferred scheme of the information management method based on the electronic bidding transaction platform, the specific implementation process of step S5 comprises:

[0030] Based on the graph structure and the risk matrix, the risk strength score is mapped to the corresponding graph edge weight to generate a weighted risk graph, wherein the color and thickness of the edge reflect the risk level;

[0031] The node clustering algorithm is applied to the weighted graph, and the bidding units in the risk edge weight concentrated area are aggregated and displayed to generate a risk cluster area, and the center unit and the average risk value are output for each subgraph.

[0032] As a preferred scheme of the information management method based on the electronic bidding transaction platform, the specific implementation process of step S5 further comprises:

[0033] In each risk cluster area, the behavior trajectory path is labeled, including the risk shortest path from the unit node to other nodes, and the semantic behavior index is labeled in the path to form a graph trajectory view;

[0034] An artificial review interface is introduced in the graph view interface, and the risk cluster and its path are labeled by the supervisor, and the label is written back to the structure graph and the risk matrix is updated;

[0035] The graph trajectory view further comprises a risk progression model driven by a causal rule chain, the risk progression model is based on the explicit or implicit relationship between the bidding units to construct a plurality of causal rule chains based on behavior and semantic signals, each causal rule chain is composed of a group of behavior events with time sequence and deduction relationship;

[0036] The risk evolution path is quantitatively modeled based on the risk progression model, a causal rule chain score is calculated, a risk level cumulative increment between bidding units is calculated based on the causal rule chain score, a preset risk label cumulative increase threshold is set, if a risk label cumulative increase between any two bidding units is greater than or equal to the risk label cumulative increase threshold, the any two bidding units are marked as a risk escalation state, and the risk level cumulative increment and the risk correlation strength score are jointly calculated to generate an updated risk level label between the bidding units.

[0037] Based on the updated risk level label, an edge weight between the bidding units in the graph trajectory view is updated to a risk strength corresponding to the updated risk level label.

[0038] In a second aspect, an embodiment of the present application provides an information management system based on an electronic bidding transaction platform, which includes an information collection module for collecting business registration information, historical bidding records, project participation conditions and related contract information of bidding units from the electronic bidding platform, and performing standardized and structured processing on the data to establish a basic information node; a graph construction module for constructing a graph structure between the bidding units based on the structured data, using rule reasoning and entity matching technology, including explicit or implicit correlation relationships, the explicit or implicit correlation relationships including legal association, equity penetration, historical consortium and project cooperation; a semantic analysis module for performing semantic feature extraction and similarity comparison on the bidding documents of the current project using natural language processing technology, identifying whether there is a suspicious behavior pattern among the units, the suspicious behavior pattern including highly similar document content, term repetition and templated structure; a risk modeling module for modeling the structural correlation information and the semantic behavior analysis result, calculating the risk correlation strength between the bidding units, and generating a risk level label or a warning prompt; a visual decision module for presenting the risk correlation strength and the risk level label in a graph visualization manner, supporting a supervisor to view the correlation path and behavior trajectory of the bidding units, and providing an artificial review interface for auxiliary judgment, forming a closed-loop processing mechanism.

[0039] In a third aspect, an embodiment of the present application provides a computer device including a memory and a processor, the memory storing a computer program, wherein the computer program instructions are executed by the processor to implement the steps of the information management method based on the electronic bidding transaction platform according to the first aspect of the present application.

[0040] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program instructions are executed by the processor to implement the steps of the information management method based on the electronic bidding transaction platform according to the first aspect of the present application.

[0041] The beneficial effects of this invention are as follows: By constructing a multi-dimensional, multi-layered information management system, this invention achieves in-depth mining and intelligent identification of potential relationships and behavioral patterns among bidding entities, thereby effectively improving the electronic bidding platform's capabilities in risk prevention and abnormal behavior identification. By integrating multi-source heterogeneous data such as business registration information, bidding history, and project contracts, and introducing graph modeling and semantic analysis technologies, this invention can accurately identify key risk factors such as corporate penetration, historical cooperation, and highly similar bid documents. It then quantifies and assesses the strength of risk associations among bidding entities, forming a visual graph and an auxiliary review mechanism, thus improving the judgment efficiency of regulatory personnel.

[0042] This invention has good scalability and operability in practical applications. Through standardized and structured data processing, it reduces the risk of information inconsistency interfering with the analysis results. At the same time, the fusion model realizes early identification of potential bid rigging and collusion in the cross-comparison of structural paths and semantic features, effectively preventing illegal bidding from undermining market fairness.

[0043] In summary, this invention not only improves the electronic bidding platform's ability to automatically identify risky behaviors, but also provides important support for building a transparent, trustworthy, and regulated digital transaction environment. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Fig. 1 This is a schematic diagram illustrating the steps of an information management method based on an electronic bidding and tendering platform.

[0046] Fig. 2 This is a schematic diagram of the structure of an information management system based on an electronic bidding and tendering platform. Detailed Implementation

[0047] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0048] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0049] Second, the "one embodiment" or "an embodiment" referred to herein means a specific feature, structure, characteristic, or combination of features and / or characteristics described herein that can be included in at least one implementation of the present application. The various appearances of "in one embodiment" or "an embodiment" in the specification are not necessarily all referring to the same embodiment.

[0050] Referring to Figs. 1-2 For one embodiment of the present application, the embodiment provides an information management method based on an electronic bidding transaction platform, comprising,

[0051] Step S1: Collecting bidding unit data information from the electronic bidding platform, and performing standardized and structured processing on the data information to establish a basic data information node.

[0052] Specifically, the original information data set of the bidding unit on the electronic bidding platform is collected, the original information data set includes business registration information, historical bidding records, project bidding information and contract clause text, and an identification field is set according to data source, data timestamp and data type for the original information data set;

[0053] A preset field template is set, and the unstructured fields in the original information data set are reorganized and regularized, the legal person information, shareholder structure, historical project keywords and contract clause abstract are reconstructed into a structured field set, and a field confidence parameter is marked; the preset field template is set according to the frequency statistics results of the structured fields of the platform historical bidding information, and the field template set is formed in combination with standard data items of different industries (such as bidding document, contract document structure). The template content includes but is not limited to the name of the legal person, the unified social credit code, the bidding response segment field, the contract clause field, etc.; the historical project keywords refer to words or phrases that can represent the core characteristics of past bidding projects, which are usually extracted from the project name, project description and bidding scope through the TF-IDF algorithm.

[0054] Further, the semantic uniform operation is performed on the natural language content in the structured field set, the contract clause text, the project bidding information and the service clause in the contract clause text are standardized processed based on the semantic normalization rule, and a uniform semantic expression field set is obtained;

[0055] Based on the structured field set and the uniform semantic expression field set, a bidding unit basic node is established, the bidding unit basic node contains an identification field, a field confidence parameter and a semantic expression field, and serves as the basic metadata of the unit node in the knowledge graph.

[0056] Step S2: Based on the structured data information, a graph structure between bidding units is constructed using rule reasoning and entity matching technology, covering explicit or implicit association relationships.

[0057] Specifically, the legal person name field, contact information field and registered address field in the basic information node are extracted to form a candidate entity set, and entities with a field confidence greater than a preset entity confidence threshold are taken as a high-confidence entity set; the entity confidence threshold is set according to the confidence distribution of the historical annotation data verified by manual verification as "real matching entity", and the statistical mean plus one standard deviation is used as the initial threshold value, and the optimal threshold value can also be automatically adjusted through cross-validation.

[0058] Further, according to the legal name field and registered capital field in the high-confidence entity set, the control path relationship strength between two bidding units is calculated under the preset equity reasoning rule condition, and the following formula is referred to:

[0059] ;

[0060] Wherein, represents the control path relationship strength between the i-th bidding unit and the j-th bidding unit, represents the influence of the intermediate node in the k-th path in the equity or position relationship, which is set according to the equity proportion or position level (such as when the holding proportion is greater than or equal to 50%, 1; when the holding proportion is between 20% and 50%, 0.5), represents the contribution value of the k-th path between the i-th bidding unit and the j-th bidding unit in the path (combined with industry expert experience, based on type, strength, and legal binding degree score), represents the shortest path length between the i-th bidding unit and the j-th bidding unit, and K represents the total number of paths.

[0061] If the control path relationship strength between the two bidding units is greater than or equal to the preset control path relationship strength threshold, the two bidding unit basic nodes are connected as equity control relationship edges;

[0062] The historical bidding project keyword field and the bid project time field in the high-confidence entity set are compared, and according to the project participation time overlap rate and keyword semantic similarity, it is determined whether the two bidding units have a historical joint bidding behavior, and a joint behavior relationship edge is constructed, and the following formula is referred to:

[0063] ;

[0064] denotes the historical joint bidding behavior score between the i-th bidding unit and the j-th bidding unit, denotes the historical bidding project time set of the i-th bidding unit, denotes the historical bidding project time set of the j-th bidding unit, denotes the historical bidding project keyword set of the i-th bidding unit, denotes the past bidding project keyword set of the j-th bidding unit, which refers to a set of high-frequency industry terms extracted from historical bidding projects, such as "municipal infrastructure", "survey and design", "informationization operation and maintenance", "underground pipe gallery", "new energy photovoltaic construction", etc., and the keyword extraction algorithm (such as TextRank, TF-IDF) is used to automatically obtain the keywords from the bidding announcement text.

[0065] Further, the graph structure composed of unit nodes and stock ownership control relationship edges and joint behavior relationship edges is taken as the initial knowledge graph between bidding units, and the node inherits the identification field and the semantic expression field in the graph structure.

[0066] Step S3: using natural language processing technology to extract semantic features and similarity comparison on the bidding documents of the current project, and identifying multiple suspicious behavior patterns of units.

[0067] Specifically, text segmentation processing is performed on each document in the current project bidding document set, a set of semantic substructures composed of bidding clause response segments, technical parameter segments, and enterprise introduction segments is extracted, and each semantic segment is labeled with paragraph position number and original unit.

[0068] Based on the set of semantic substructures, a pre-trained language model (such as BERT or RoBERTa) is used to perform word segmentation and stop word removal on each semantic segment, and the pre-trained language model is used to obtain sentence vectors (taking the vector corresponding to the [CLS] label, such as a 768-dimensional vector), and the sentence vectors are subjected to L2 normalization processing; according to the set window length, the semantic segment vectors submitted by different bidding units are compared two by two to construct a semantic comparison matrix;

[0069] Further, all elements in the semantic comparison matrix are evaluated, and if there are three or more semantic segments with similarity higher than the preset semantic consistency threshold between any bidding unit basic node pairs, they are marked as a set of highly consistent semantics;

[0070] For the semantic structure matching of all unit pairs in the set of highly consistent semantics, the semantic repetition index is calculated according to the distribution density of similar paragraphs and the proportion of template structure, and the following formula is referred to:

[0071] ;

[0072] wherein, represents the semantic repetition degree between the i-th bidding unit and the j-th bidding unit in the bidding documents, represents the total number of terms in the file, represents the word vector of the m-th term in the bidding documents of the i-th bidding unit, represents the word vector of the m-th term in the bidding documents of the j-th bidding unit.

[0073] Step S4: fuse the structural association information with the semantic behavior analysis results, calculate the risk association strength between the bidding units, and generate risk level labels or warning prompts.

[0074] Specifically, based on the path set between any two bidding unit basic nodes corresponding to the units in the initial knowledge graph between the bidding units, the path length, path type distribution and associated node density are extracted to form a structural association vector;

[0075] The semantic repetition degree index of any two bidding unit basic nodes corresponding to the units is called, and a semantic behavior vector is constructed, including semantic repetition density and template structure proportion;

[0076] Further, a fusion model is constructed to calculate the risk association strength value of any two bidding unit basic nodes corresponding to the units, the input of the fusion model is the structural association vector and the semantic behavior vector, the output is the risk strength score between 0 and 1, and the risk strength threshold value is set, the risk level label is generated according to the score, and the following formula is referred to:

[0077] ;

[0078] wherein, represents the risk association strength score between the i-th bidding unit and the j-th bidding unit, is a sigmoid function used to map the score to the interval of 0~1, and are the weight coefficients of the structure and behavior vectors, respectively, is a structure vector, representing the feature combination of graph paths and historical behavior relationships, is a behavior vector, representing suspicious behavior features such as semantic repetition.

[0079] The risk level division rules are as follows:

[0080] ;

[0081] wherein, represents the risk level label between the i-th bidding unit and the j-th bidding unit, represents a pre-set high-risk threshold, represents a pre-set medium-risk threshold.

[0082] All risk intensity scores are aggregated to form an inter-unit risk matrix.

[0083] Step S5: The risk association strength and risk level label are presented in a graph visualization manner, supporting the supervisor to view the association path and behavior trajectory of the bidding unit, and providing an artificial review interface to assist in judgment, forming a closed-loop processing mechanism.

[0084] Specifically, based on the graph structure and the risk matrix, the risk intensity score is mapped to the corresponding graph edge weight to generate a weighted risk graph, wherein the color and thickness of the edge reflect the risk level.

[0085] The node clustering algorithm is applied to the weighted graph to aggregate and display the bidding units in the risk edge weight concentrated area, generate a risk cluster area, and output the center unit and the average risk value for each subgraph.

[0086] Further, the behavior trajectory path is marked in each risk cluster area, including the risk shortest path from the unit node to other nodes, and the semantic behavior indicators are marked in the path to form a graph trajectory view.

[0087] An artificial review interface is introduced in the graph view interface, and the supervisor marks the risk cluster and its path, including "association confirmation", "non-association", "to be reviewed" and other label states, which are written back to the structure graph and update the risk matrix.

[0088] The graph trajectory view also includes a risk progression model driven by a chain of causal rules, which is based on the explicit or implicit relationship between bidding units to construct a number of causal rule chains based on behavior and semantic signals. Each causal rule chain is composed of a group of behavior events with temporal and deductive relationships.

[0089] Based on the risk progression model, the risk evolution path is quantitatively modeled, and the causal rule chain score is calculated, as follows:

[0090] ;

[0091] Wherein, represents the causal rule chain score of the qth causal rule chain acting on the i th bidding unit and the j th bidding unit, represents the weight coefficient of the h th causal event (obtained according to expert experience or data training), represents the judgment function, if the i th bidding unit and the j th bidding unit trigger the h th causal event, then takes the value of 1, otherwise 0, represents the number of causal events in the q th causal rule chain.

[0092] Based on the causal rule chain score, the risk level cumulative increment between the bidding units is calculated, and the formula is as follows:

[0093] ;

[0094] Wherein, denotes the risk level cumulative increment between the i th bidding unit and the j th bidding unit, denotes a preset risk promotion step coefficient, denotes the total number of causal rule chains, denotes a preset triggering threshold value of the q th causal rule chain, denotes an indicator function, if 1, otherwise 0.

[0095] A preset risk label cumulative increase threshold value, if the risk label cumulative increase between any two bidding units is greater than or equal to the risk label cumulative increase threshold value, any two bidding units are marked as a risk escalation state, and the risk level cumulative increment and the risk association intensity score are jointly calculated to generate an updated risk level label between the bidding units, and the calculation formula is as follows: Wherein, The updated risk level label between the i th bidding unit and the j th bidding unit.

[0096] Based on the updated risk level label, the edge weight between the bidding units is updated to the risk intensity corresponding to the updated risk level label in the graph trajectory view.

[0097] Further, the embodiment also provides an information management system based on an electronic bidding transaction platform, comprising: an information collection module, configured to collect business registration information, historical bidding records, project participation conditions and related contract information of bidding units from the electronic bidding platform, and perform standardization and structurization processing on the data to establish a basic information node; a graph construction module, configured to construct a graph structure between the bidding units based on the structured data, using rule reasoning and entity matching technology, including explicit or implicit correlation, wherein the explicit or implicit correlation includes legal association, equity penetration, historical consortium and project cooperation; a semantic analysis module, configured to perform semantic feature extraction and similarity comparison on a bidding document of a current project using natural language processing technology, and identify whether there is a suspicious behavior pattern among the units, wherein the suspicious behavior pattern includes highly similar document content, repeated terms and templated structure; a risk modeling module, configured to fuse and model the structural correlation information and the semantic behavior analysis result, calculate the risk correlation strength between the bidding units, and generate a risk level label or a warning prompt; and a visual decision module, configured to present the risk correlation strength and the risk level label in a graph visualization manner, support a supervisor to view the correlation path and behavior track of the bidding units, and provide an artificial review interface for auxiliary judgment, forming a closed-loop processing mechanism.

[0098] The embodiment also provides a computer device suitable for the information management method based on the electronic bidding transaction platform, comprising a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions to implement the information management method based on the electronic bidding transaction platform.

[0099] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or can be a key, trackball or touchpad arranged on the shell of the computer device. In addition, the input device can be an external keyboard, touchpad or mouse, etc.

[0100] The embodiment also provides a storage medium, which stores a computer program, and the computer program is executed by a processor to implement the information management method based on the electronic bidding transaction platform.

[0101] To sum up, the application realizes in-depth mining and intelligent identification of potential correlation between bidding units and behavior patterns by constructing a multi-dimensional and multi-level information management system, thereby effectively improving the ability of the electronic bidding platform in risk prevention and control and abnormal behavior identification. By fusing multi-source heterogeneous data such as business registration information, bidding history, project contract, and introducing graph modeling and semantic analysis technology, the application can accurately identify key risk factors such as corporate penetration, historical cooperation, and high similarity of bidding documents, and then quantitatively evaluate the risk correlation strength between bidding units, and form a visual graph and an auxiliary review mechanism, thereby improving the judgment efficiency of supervisors.

[0102] The application has good expansibility and operability in practical application. Through a standardized and structured data processing process, the interference risk of information inconsistency on the analysis result is reduced. At the same time, the fusion model realizes early identification of potential serial bidding and bid-rigging behavior in the cross comparison of structural paths and semantic features, effectively preventing the destruction of market fairness by irregular bidding.

[0103] To sum up, the application not only improves the automatic identification ability of the electronic bidding platform for risk behavior, but also provides important support for building a transparent, credible, and monitorable digital transaction environment.

[0104] It should be noted that the above embodiments are only used to illustrate the technical solutions of the application and are not limiting. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the application, and they should be covered in the scope of the claims of the application.

Claims

1. An information management method based on an electronic bidding transaction platform, characterized in that: include: Step S1: Collect bidding unit data from the electronic bidding platform, standardize and structure the data, and establish basic data information nodes; Step S2: Based on the structured data information, construct the graph structure between bidding units using rule reasoning and entity matching technology, including explicit or implicit relationships; Step S3: Use natural language processing technology to extract semantic features and compare similarity in the tender documents of the current project to identify suspicious behavior patterns of multiple entities; Step S4: Integrate the structural association information with the semantic behavior analysis results to form a model, calculate the risk association strength between bidding units, and generate risk level labels; Step S5: Present the risk association strength and risk level labels in a graphical visualization, allowing regulators to view the association paths and behavioral trajectories of bidding units, and providing a manual review interface to assist in judgment, forming a closed-loop processing mechanism; Step S1 includes: Collect the original information dataset of bidding entities on the electronic bidding platform. The original information dataset includes business registration information, historical bidding records, project winning information and contract terms text. Set identification fields for the original information dataset according to data source, data timestamp and data type. Preset field templates and reorganize and standardize the unstructured fields in the original information dataset to reconstruct the legal person information, shareholder structure, historical project keywords and contract clause summaries into a set of structured fields, and label the field confidence parameters; Semantic unification is performed on the natural language content in the structured field set. Based on the semantic normalization rule, the contract terms text, project winning information, and service terms in the contract terms text are standardized to obtain a unified semantic expression field set. Based on the structured field set and the unified semantic expression field set, a basic node for bidding units is established. The basic node for bidding units includes an identifier field, a field confidence parameter, and a semantic expression field, and serves as the basic metadata for unit nodes in the knowledge graph. Step S2 includes: Extract the legal entity name field, contact information field, and registered address field from the basic information node to form a candidate entity set, and use the entities whose field confidence scores are greater than the preset entity confidence score threshold as the high-confidence entity set; Based on the legal person name field and registered capital field in the set of highly trustworthy entities, under the preset equity inference rules, the control path relationship strength between the two bidding entities is calculated. If the control path relationship strength between the two bidding entities is greater than or equal to the preset control path relationship strength threshold, the basic nodes of the two bidding entities are connected as equity control relationship edges. The keyword field of historical bidding projects and the time field of winning projects in the set of highly trustworthy entities are jointly compared. Based on the overlap rate of project participation time and the semantic similarity of keywords, it is determined whether there is historical joint bidding behavior between the two bidding entities, and joint behavior relationship edge is constructed. The graph structure composed of unit nodes, equity control relationship edges, and joint behavior relationship edges is used as the initial knowledge graph among bidding units. In the graph structure, nodes inherit the identifier field and the semantic expression field. Step S3 includes: Perform text segmentation processing on each document in the current project tender document set, extract a semantic substructure set consisting of the tender clause response segment, technical parameter segment, and company introduction segment, and mark each semantic segment with paragraph position number and original unit; Based on the set of semantic substructures, a context embedding model is used to generate a vector representation of each semantic segment. The semantic segment vectors submitted by different bidding units are compared pairwise according to the set window length to construct a semantic comparison matrix. All elements in the semantic comparison matrix are evaluated. If any pair of units has three or more semantic segments with similarity higher than the preset semantic consistency threshold, it is marked as a set with highly consistent semantics. For the semantic structure matching of all unit pairs in the semantically highly consistent set, a semantic repetition index is calculated based on the distribution density of similar paragraphs and the proportion of template structure.

2. The information management method based on the electronic bidding transaction platform according to claim 1, characterized in that, Step S4 includes: Based on the set of paths between any two basic nodes of bidding units in the initial knowledge graph among bidding units, the path length, path type distribution and associated node density are extracted to form a structural association vector. Retrieve the semantic repetition index of the corresponding units of any two bidding units' basic nodes, and construct a semantic behavior vector, including semantic repetition density and template structure ratio; Construct a fusion model to calculate the risk association strength value of the corresponding units of any two bidding units' basic nodes. The input of the fusion model is the structural association vector and the semantic behavior vector, and the output is the normalized risk strength score. Set a risk strength threshold and generate a risk level label based on the score. All risk intensity scores are aggregated to form an inter-unit risk matrix.

3. The information management method based on the electronic bidding transaction platform according to claim 2, characterized in that, Step S5 includes: Based on the graph structure and risk matrix, the risk intensity score is mapped to the corresponding graph edge weight to generate a weighted risk graph, where the color and thickness of the edges reflect the risk level. A node clustering algorithm is applied to the weighted graph to aggregate and display bidding units in areas with concentrated risk edge weights, generating risk cluster areas, and outputting the central unit and average risk value for each subgraph.

4. The information management method based on the electronic bidding transaction platform according to claim 3, characterized in that, Step S5 further includes: In each risk cluster region, the behavioral trajectory path is marked, including the shortest risk path from the unit node to other nodes. Semantic behavioral indicators are marked in the path to form a graph trajectory view. A manual review interface is introduced into the graph view interface, where regulatory personnel can mark risk clusters and their paths, including the status of associated confirmation, non-associated, and pending review. The labels will be written back to the structural graph and the risk matrix will be updated. The graph trajectory view also includes a risk progression model driven by causal rule chains. The risk progression model constructs several causal rule chains based on behavioral and semantic signals based on the explicit or implicit relationships between bidding units. Each causal rule chain consists of a set of behavioral events with temporal sequence and inferential relationship. Based on the risk progression model, the risk evolution path is quantitatively modeled, and the causal rule chain score is calculated. Based on the causal rule chain score, the cumulative increment of risk level between bidding units is calculated. A cumulative increase threshold for risk label is preset. If the cumulative increase of risk label between any two bidding units is greater than or equal to the cumulative increase threshold for risk label, then any two bidding units are marked as risk escalation status. The cumulative increment of risk level is then jointly calculated with the risk correlation strength score to generate updated risk level labels between bidding units. Based on the updated risk level label, the edge weights between bidding entities in the map trajectory view are updated to the risk intensity corresponding to the updated risk level label.

5. An information management system based on an electronic bidding transaction platform, based on the information management method of any one of claims 1-4, characterized in that: Also includes: The information collection module is used to collect the business registration information, historical bidding records, project participation information and related contract information of bidding units from the electronic bidding platform, and to standardize and structure the data to establish basic information nodes. The graph construction module is used to construct a graph structure between bidding units based on structured data and using rule reasoning and entity matching technology, including explicit or implicit relationships. The explicit or implicit relationships include legal person relationships, equity penetration, historical alliances and project cooperation. The semantic analysis module is used to extract semantic features and compare similarities in the tender documents of the current project using natural language processing technology, and to identify whether multiple entities have suspicious behavior patterns. The suspicious behavior patterns include highly similar document content, repeated terms, and templated structures. The risk modeling module is used to integrate structural correlation information with semantic behavior analysis results to calculate the risk correlation strength between bidding units and generate risk level labels or early warning prompts. The visual decision-making module is used to present the risk correlation strength and risk level labels in a graphical visualization manner, allowing regulators to view the correlation paths and behavioral trajectories of bidding units, and providing a manual review interface to assist in judgment, forming a closed-loop processing mechanism. 6.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: When the processor executes the computer program, it implements the steps of the information management method based on an electronic bidding and tendering transaction platform as described in any one of claims 1 to 4.

7. A computer readable storage medium having stored thereon a computer program, characterized in that: When the computer program is executed by the processor, it implements the steps of the information management method based on an electronic bidding and tendering transaction platform as described in any one of claims 1 to 4.

Images