Road transportation market illegal behavior situation awareness system and method thereof

Abstract

The invention relates to the technical field of road transportation informatization systems, in particular to a road transportation market illegal behavior situation awareness system and a road transportation market illegal behavior situation awareness method. The collected data is preprocessed and stored; the relation matrix construction module is in communication connection with the data acquisition module; based on the preprocessed data, constructing a market subject relation matrix R; the propagation intensity calculation module is in communication connection with the relation matrix construction module; calculating an illegal behavior propagation intensity matrix P; the topology network construction module is in communication connection with the propagation intensity calculation module; constructing a topology propagation network G; the behavior evolution prediction module is in communication connection with the topology network construction module; based on the topology propagation network G, predicting the evolution of the violation behavior; the risk assessment module is in communication connection with the behavior evolution prediction module; and risk assessment is performed and early warning information is generated based on the illegal behavior evolution prediction result, so that a potential wind source can be identified as early as possible, and the early warning timeliness is greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of road transport information systems, and in particular to a road transport market illegal behavior situation awareness system and a method thereof. Background Art

[0002] In recent years, with the rapid development of the road transport market, the market scale has continued to expand, the number of participants has increased, and the market structure has become increasingly complex. At the same time, market violations have also shown a trend of diversification and concealment, posing a huge challenge to market supervision. Traditional supervision methods mainly rely on manual inspections and simple data analysis, which are difficult to cope with the current complex and changing market environment.

[0003] Currently, the industry has begun to try to use big data and artificial intelligence technologies to improve regulatory efficiency. For example, some methods use social network analysis to identify the relationship between market players to predict potential violations. Other methods use time series analysis to predict the violation trend of a single market player. However, these existing technologies still have many shortcomings.

[0004] First, most existing methods analyze each market player in isolation, ignoring the complex interactions between them. In reality, the violation of one player often affects other players associated with it, forming a contagion effect. Existing technologies have difficulty capturing this complex transmission dynamic.

[0005] Secondly, existing methods usually only focus on one aspect of data, such as only analyzing transaction data or only considering historical violation records. This single-dimensional analysis cannot fully reflect the true situation of the market and is prone to missing important risk signals.

[0006] Furthermore, many existing technologies lack the ability to adapt to the dynamic changes in the market environment. The road transport market is a rapidly changing environment, with new market players constantly joining and old players possibly exiting, and market relations also continuously evolving. Static analysis models are unable to capture these changes in a timely manner, resulting in a significant reduction in the timeliness and accuracy of the forecast results.

[0007] In addition, the existing early warning systems often have a high false alarm rate and a short warning time. This not only wastes regulatory resources, but also weakens the regulatory authorities’ trust in early warning information. At the same time, insufficient warning time also makes it difficult for regulatory authorities to take timely and effective intervention measures.

[0008] Finally, existing technologies often use simple scoring models in risk assessment, which makes it difficult to fully consider the combined impact of the current status, changing trends and surrounding environment. This crude risk assessment method is difficult to provide strong support for precise supervision. Summary of the invention

[0009] In response to the above problems, the present invention proposes an innovative road transport market illegal behavior situation awareness system and method. The system aims to comprehensively improve the perception and prediction capabilities of market illegal behaviors through innovative technologies such as multi-source data fusion, complex network analysis, dynamic propagation model and intelligent risk assessment.

[0010] The present invention proposes a road transport market illegal behavior situation awareness system, comprising: Data acquisition module for: Collect online public opinion, government regulatory data, and market entity transaction data; Preprocess and store the collected data; A relationship matrix building module is connected in communication with the data acquisition module and is used to: Receiving the preprocessed data sent by the data acquisition module; Based on the preprocessed data, a market subject relationship matrix R is constructed; The propagation intensity calculation module is communicatively connected with the relationship matrix construction module and is used for: Receive the market entity relationship matrix R sent by the relationship matrix building module; Based on the market subject relationship matrix R, the violation propagation intensity matrix P is calculated; A topology network construction module is communicatively connected with the propagation intensity calculation module and is used for: Receiving the violation propagation intensity matrix P sent by the propagation intensity calculation module; Based on the violation propagation intensity matrix P, a topological propagation network G is constructed; The behavior evolution prediction module is connected to the topology network construction module for: Receiving the topology propagation network G sent by the topology network construction module; Based on the topological propagation network G, predict the evolution of illegal behaviors; A risk assessment module is connected in communication with the behavior evolution prediction module and is used to: Receiving the illegal behavior evolution prediction result sent by the behavior evolution prediction module; Based on the predicted results of the evolution of the illegal behavior, risk assessment is performed and early warning information is generated.

[0011] Preferably, the relationship matrix building module comprises: A relationship factor extraction unit, used to extract a competitive relationship index, a transaction frequency index and a spatial proximity index from the preprocessed data; A weight calculation unit, used to calculate the weight coefficients of the competition relationship index, the transaction frequency index and the spatial proximity index; The matrix generating unit is used to generate a market subject relationship matrix R based on the index extracted by the relationship factor extracting unit and the weight coefficient calculated by the weight calculating unit.

[0012] Preferably, the propagation intensity calculation module comprises: A historical violation matrix generating unit, used for generating a historical violation matrix H based on the preprocessed data; The Hadamard product calculation unit is used to calculate the Hadamard product of the market entity relationship matrix R and the historical violation matrix H to obtain the violation propagation intensity matrix P.

[0013] Preferably, the topology network building module includes: A threshold setting unit, used for setting a propagation threshold θ; An adjacency matrix generating unit, used for generating an adjacency matrix A1 based on the propagation threshold θ and the violation propagation intensity matrix P; A graph structure building unit is used to build a topological propagation network G based on the adjacency matrix A1.

[0014] Preferably, the behavior evolution prediction module comprises: A state vector initialization unit, used to initialize the violation state vector S(t) of the market entity; An evolution parameter setting unit, used to set the propagation rate λ and the recovery rate μ; The state updating unit is used to calculate the updated value S(t+1) of the state vector based on the topological propagation network G, the state vector S(t), the propagation rate λ and the recovery rate μ.

[0015] Preferably, the risk assessment module comprises: A weight setting unit is used to set the weight coefficient of the current state, change trend and neighbor influence; A risk calculation unit, used to calculate the risk value of each market entity based on the weight coefficient and the prediction result of the evolution of the illegal behavior; The warning generation unit is used to generate warning information according to the calculated risk value and the preset warning threshold.

[0016] Preferably, the data collection module collects network public opinion data from the Internet through web crawler technology, obtains regulatory data from government regulatory departments through an API interface, and obtains market entity transaction data from a third-party payment company through a data exchange platform.

[0017] Preferably, a data preprocessing module is also included for cleaning, deduplicating, and formatting the collected raw data, and performing word segmentation and feature extraction on the text data using natural language processing technology.

[0018] Preferably, a visual display module is also included for visually displaying the risk assessment results in the form of charts and heat maps, and supporting interactive query and analysis.

[0019] The method for situation awareness of illegal behaviors in the road transport market based on the system comprises the following steps: S1: Collect online public opinion, government supervision data and market entity transaction data, and perform pre-processing; S2: Based on the preprocessed data, construct the market entity relationship matrix R; S3: Based on the market subject relationship matrix R, calculate the violation propagation intensity matrix P; S4: Based on the violation propagation intensity matrix P, a topological propagation network G is constructed; S5: Predict the evolution of violation behaviors based on the topological propagation network G; S6: Based on the prediction results of the evolution of illegal behaviors, conduct risk assessment and generate early warning information.

[0020] The beneficial effects of the present invention are mainly reflected in the following aspects: The road transport market illegal behavior situation awareness system and method of the present invention have significant technical advantages and practical application value. First, by constructing a multi-dimensional market subject relationship matrix, the system can comprehensively capture the complex relationship between market subjects, laying a solid foundation for subsequent analysis. This method not only takes into account direct transaction relationships, but also incorporates factors such as competitive relationships and geographical locations, greatly improving the depth of understanding of market structure.

[0021] Secondly, the present invention innovatively introduces an improved SIR propagation model, combining the propagation process of illegal behavior with complex network theory. This method can accurately simulate the diffusion process of illegal behavior in the market and reveal the propagation law of illegal behavior. Through this dynamic model, the system can identify potential risk sources early, greatly improving the timeliness of early warning.

[0022] Furthermore, the topological network construction method of the present invention cleverly transforms the continuous propagation intensity into a discrete network structure, which not only retains key information but also reduces the computational complexity. This method enables the system to efficiently process large-scale market data and achieve real-time analysis and prediction.

[0023] In addition, the comprehensive risk assessment method proposed in this invention realizes a comprehensive assessment of the risks of market entities by simultaneously considering the current status, change trend and neighbor influence. This multi-dimensional risk assessment method greatly improves the accuracy of risk identification, effectively reduces the false alarm rate, and provides a more reliable decision-making basis for regulatory authorities.

[0024] Finally, the system of the present invention has good adaptability and scalability. By adjusting key parameters, the system can adapt to different types of market environments and illegal behavior patterns. This flexibility enables the system to be continuously optimized and improved in practical applications, providing continuous technical support for long-term market supervision.

[0025] In general, the present invention effectively solves the problems existing in the prior art through innovative technical solutions, and significantly improves the accuracy, timeliness and reliability of illegal behavior prediction. This can not only help regulatory authorities to identify market risks more accurately, but also provide strong support for the formulation of targeted regulatory strategies, thereby promoting the healthy and orderly development of the road transport market. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is the overall logic block diagram of the system of the present invention.

[0027] Figure 2 It is a logic block diagram of the data acquisition module of the present invention.

[0028] Figure 3 It is a logic block diagram of the relationship matrix building module of the present invention.

[0029] Figure 4 It is a logic block diagram of the propagation intensity calculation module of the present invention.

[0030] Figure 5 It is a logical block diagram of the topological network building module of the present invention.

[0031] Figure 6 It is a logic block diagram of the behavior evolution prediction module of the present invention.

[0032] Figure 7 It is a logic block diagram of the risk assessment module of the present invention. DETAILED DESCRIPTION

[0033] See also Figure 1-7 The present invention provides a road transport market illegal behavior situation awareness system and method. The system includes multiple functional modules, and aims to achieve comprehensive awareness and early warning of illegal behaviors in the road transport market through data collection, analysis and prediction.

[0034] First, the road transport market violation situation awareness system of the present invention includes a data acquisition module 1. This module is responsible for collecting online public opinion, government supervision data and market entity transaction data, and preprocessing and storing these data. In practical applications, the data acquisition module 1 can obtain data through multiple channels. For example, network public opinion data can be obtained from social media, news websites and other platforms using web crawler technology; regulatory data can be obtained through data interfaces established with government departments; and transaction data of market entities can be obtained in cooperation with third-party payment companies. This multi-source data collection method can ensure that the system obtains comprehensive and rich information, laying the foundation for subsequent analysis.

[0035] Next, the relationship matrix construction module 2 in the system establishes a communication connection with the data acquisition module 1. The main function of the relationship matrix construction module 2 is to receive the preprocessed data and construct the market subject relationship matrix R based on the data. This process involves complex mathematical calculations and model construction. Specifically, the construction method of the market subject relationship matrix R is as follows: , in, represents the number of market players, Representative subject and subject The strength of the relationship between them. This relationship strength is calculated by the following formula: , In this formula, represents the competitive relationship index, represents the transaction frequency index, is the spatial proximity index. , β and are the corresponding weight coefficients, and their sum is equal to 1. This calculation method takes into account the multi-dimensional relationship between market entities and can more accurately reflect the actual situation.

[0036] Competitive Relationship Index Indicates the enterprise and Enterprise The intensity of competition between the two. The index can be calculated by the following steps:

[0037] 1. Market share analysis: Data source: public data such as corporate annual reports and financial statements obtained from government regulatory agencies, or market share data provided by market research companies. Calculation formula: ,in, and Respectively represent enterprises and Enterprise This formula measures the difference in market share between two businesses; the greater the difference, the more intense the competition.

[0038] 2.Pricing strategy analysis: Data source: Price data obtained from the company's quotation, contract records, and market price monitoring platform. Calculation formula: ,in, and Respectively represent enterprises and Enterprise This formula measures the similarity of two companies' pricing strategies. The closer the prices, the more intense the competition.

[0039] 3. Overlap of service scope: Data source: obtained from the company's operating areas, service outlets and other geographic information data. Calculation formula: ,in, and Respectively represent enterprises and Enterprise Area represents the service coverage area of ​​two companies. This formula measures the degree of overlap between the service areas of two companies. The more overlap, the more intense the competition.

[0040] 4. Comprehensive competitive relationship index: Calculation formula: ,in, , and is the weight coefficient, which can be adjusted according to the actual situation. Usually, the weight of market share and pricing strategy is larger, while the weight of service scope overlap is smaller.

[0041] Trading Frequency Index Indicates the enterprise and Enterprise The frequency of transactions between two parties within a certain period of time. It can be calculated by following the steps below:

[0042] 1. Data source: transaction data obtained from the company's transaction records, third-party payment platforms, logistics management systems, etc. Clean, deduplicate and format the original transaction data to ensure data accuracy and consistency.

[0043] 2. Choose the right time window: Choose the right time window based on the application scenario, such as the past year, the past six months, or the past three months. A shorter time window can capture recent changes, while a longer time window can reflect long-term trends.

[0044] 3. Transaction frequency calculation: The calculation formula is ,in, Indicates the enterprise and Enterprise The number of trades within the selected time window, represents the highest number of transactions among all enterprise pairs. This formula normalizes the transaction frequency to the interval [0, 1], making it easier to compare the transaction frequencies between different enterprise pairs.

[0045] Spatial proximity index Indicates the enterprise and Enterprise The index can be calculated by the following steps:

[0046] 1. Geographic information acquisition: Acquire from the Geographic Information System (GIS) data such as the company's registered address, warehouse location, service outlets, etc. Convert the company's geographic location into longitude and latitude coordinates for distance calculation.

[0047] 2. Distance calculation: The calculation formula is Haversine , where Haversine is a commonly used geographic distance calculation formula used to calculate the shortest spherical distance between two longitude and latitude points. and Respectively represent enterprises The latitude and longitude of and Respectively represent enterprises The latitude and longitude of .

[0048] 3. Proximity calculation: The calculation formula is ,in, is a tuning parameter used to control the effect of distance on proximity. The value makes the proximity decay rapidly with increasing distance, and the smaller the The value makes the proximity change more gradually.

[0049] In the road transport market, the relationships between different companies (such as logistics companies, freight forwarders, etc.) are complex and diverse. For example, two logistics companies may compete for business in the same area, frequently trade goods, or be geographically close, making it easy for violations to spread.

[0050] Assume there are 5 logistics companies , the relationship between them can be measured by the following three dimensions:

[0051] Competitive Relationship Index : It is obtained by analyzing market share, pricing strategy and other factors. For example, A and B are in fierce competition in the same city. , while A and C do not compete in the same city, .

[0052] Trading Frequency Index :It is directly obtained from the transaction data. For example, there is a large amount of cargo exchange between A and B every week. , while A and E rarely trade. .

[0053] Spatial proximity index :Calculated by combining geographic information system (GIS) data. For example, the warehouses of A and B are very close to each other. , while A and D are far apart. .

[0054] By comprehensively considering competition, transactions, and geographic location, the system can identify market entities that are closely related to potential violations. For example, A and B not only compete fiercely, but also have frequent transactions and are geographically close, which may be a high-risk pair for the spread of violations. , the system can provide early warning to companies that may be affected by other companies’ violations, helping regulatory authorities take timely measures.

[0055] In order to achieve the above functions, the relationship matrix construction module 2 includes several important sub-units. The first is the relationship factor extraction unit 21, which is responsible for extracting the competitive relationship index, transaction frequency index and spatial proximity index from the preprocessed data. This process may involve data mining and feature engineering techniques. For example, the competitive relationship index can be obtained by analyzing factors such as market share and price strategy; the transaction frequency index can be directly obtained from transaction data; and the spatial proximity index may need to be calculated in combination with geographic information system (GIS) data.

[0056] Secondly, the weight calculation unit 22 is used to calculate the weight coefficients of the competitive relationship index, the transaction frequency index and the spatial proximity index. The determination of these weights usually requires the combination of the experience of industry experts and the statistical analysis of historical data. A possible method is to use the analytic hierarchy process (AHP) to determine the weights, which can comprehensively consider the relative importance of multiple factors.

[0057] Finally, the matrix generation unit 23 is responsible for generating the final market subject relationship matrix R based on the extracted indexes and calculated weight coefficients. This process involves large-scale matrix operations and may require the use of high-performance computing technology to improve efficiency.

[0058] After the relationship matrix is ​​constructed, the communication intensity calculation module 3 receives the market entity relationship matrix R and further calculates the violation communication intensity matrix P. This calculation process uses an innovative mathematical model, as follows: , here, represents the Hadamard product, i.e., the element-by-element multiplication of matrices. H is the historical violation matrix, which reflects the degree of influence of past violations of market entities on other entities.

[0059] The construction of the historical violation matrix H relies on the following main data sources: Penalty records of government regulatory departments: including violation penalty notices, fine records, rectification requirements, etc. issued by transportation management departments, market supervision and administration bureaus, etc.

[0060] Enterprise credit information systems: such as the National Enterprise Credit Information Publicity System, local credit platforms, etc. These systems record information such as the company's credit rating, administrative penalties, and abnormal business lists.

[0061] Data from industry associations and third-party assessment agencies: industry self-discipline reports issued by industry associations, assessment results of corporate compliance by third-party assessment agencies, etc.

[0062] Media reports and public opinion data: reports and public feedback related to corporate violations obtained from news websites, social media and other platforms through web crawler technology.

[0063] After obtaining the raw data, the following preprocessing steps are required to ensure the quality and usability of the data:

[0064] Data cleaning: Remove duplicate, erroneous or incomplete records to ensure data accuracy.

[0065] Time standardization: Unify the time formats of different sources into a standard format (such as ISO 8601) to facilitate subsequent time series analysis.

[0066] Normalization: Classify and code different types of violations (such as overloading, speeding, illegal modifications, etc.) to ensure that different types of violations can be compared and quantified.

[0067] To construct the historical violations matrix , it is necessary to evaluate the impact of each market player’s violation on itself and other market players. The specific steps are as follows:

[0068] Define scoring criteria: Develop a scoring criteria based on the nature of the violation, the degree of harm, the severity of the penalty, etc. For example: Minor violation (such as slight speeding): 1 point; Medium violation (such as overloading): 3 points; Severe violation (such as illegal modification, unlicensed operation): 5 points; Calculate the total score: For each market entity , count all the violations in the past period of time and calculate the total score according to the scoring criteria .

[0069] If the enterprise Violations directly affected the business ,but For example, if Company A is fined for overloading, Company B, as its partner, will also be affected (such as cargo delays, customer complaints, etc.). If the value of of violations indirectly affected companies through the supply chain or other channels ,but For example, if the violation of Company A causes damage to the reputation of its supplier Company C, but the impact is small, then The value of is lower.

[0070] In order to reflect the weakening of the impact of violations on other entities over time, a time decay factor is introduced , the formula is: ,in, Indicates the difference between the time when the violation occurred and the current time. is a tuning parameter that controls the decay speed. The value makes the effect decay quickly over time, and smaller The value makes the impact change more gradually.

[0071] Comprehensive impact assessment, calculated as follows: ,in, Indicates the enterprise The number of violations in the past period of time, It is The weight of the violation (based on severity score), It is The scope of impact of the violation on other entities, It is The time when the violation occurred.

[0072] Based on the above calculation results, a historical violation matrix is ​​constructed. . Each element in the matrix Indicates the enterprise Historical violations of other companies The size of the matrix is ,in is the number of market players.

[0073] The impact of historical violations on companies cannot be ignored. For example, if a logistics company is once fined for overloading, it may cause its partners or competitors to follow suit, thus causing the spread of violations.

[0074] Assume that there is a historical violation matrix ,in Indicates past business For Enterprises For example, A was once punished for overloading, which had a negative impact on B. , but has almost no effect on C. .

[0075] By converting the current relationship strength matrix Matrix with historical violations Combined with the above, the system can evaluate the impact of historical violations on the current market. For example, if the relationship between A and B is strong and A has a history of violations, this means that B also has a high risk of violations.

[0076] The Hadamard product operation enables the system to accurately predict the propagation path of violations. For example, if the relationship strength between A and B is 0.8, and A has a history of violations ,So , indicating that B is more affected by A’s violation.

[0077] To achieve this calculation process, the propagation intensity calculation module 3 includes two key sub-units. First, the historical violation matrix generation unit 31 is responsible for generating the historical violation matrix H based on the preprocessed data. This process may involve time series analysis and pattern recognition techniques to extract meaningful violation patterns from historical data.

[0078] Secondly, the Hadamard product calculation unit 32 performs the Hadamard product operation of the market subject relationship matrix R and the historical violation matrix H, and finally obtains the violation propagation intensity matrix P. This calculation method can effectively combine the current market relationship and historical violation, so as to more accurately predict the propagation trend of the violation.

[0079] From the above description, it can be seen that the system of the present invention adopts innovative methods and technologies in data collection, relationship matrix construction and propagation intensity calculation. These innovations not only improve the accuracy and efficiency of the system, but also adapt to the complex and changing road transport market environment. For example, by introducing the spatial proximity index, the system can better capture the impact of geographical location on the propagation of violations; and by considering historical violations, the system can identify market entities with a tendency to repeat violations.

[0080] This multi-dimensional and multi-level analysis method enables the system of the present invention to understand the illegal behavior situation in the road transport market more comprehensively and deeply. It can not only identify current illegal behaviors, but also predict the illegal trends that may appear in the future, providing timely and effective decision-making support for regulatory authorities.

[0081] In general, the road transport market illegal behavior situation awareness system of the present invention provides a new idea and tool for solving the problem of illegal behavior in the road transport market through innovative data processing and analysis methods. Its application is expected to significantly improve the efficiency and accuracy of market supervision, thereby promoting the healthy and orderly development of the road transport market. The road transport market illegal behavior situation awareness system of the present invention is further explored to elaborate on other key modules and their functions in the system.

[0082] After the propagation intensity calculation is completed, the system of the present invention introduces a topological network construction module 4. This module establishes a communication connection with the propagation intensity calculation module 3, and its main task is to receive the violation propagation intensity matrix P and construct a topological propagation network G based on it. The introduction of this step reflects the innovation of the present invention in the application of network science.

[0083] Specifically, the topological network construction module 4 contains three subunits, each of which plays an important role in the network construction process. First, the threshold setting unit 41 is responsible for setting the propagation threshold θ. The choice of this threshold is crucial, as it determines which relationships between market entities will be included in the network structure. In practical applications, the choice of thresholds may need to be adjusted according to the specific market environment and regulatory requirements. For example, in a highly competitive market, a lower threshold may need to be set to capture more potential associations; whereas in a relatively stable market, a higher threshold may be required to focus on the most significant relationships.

[0084] Secondly, the adjacency matrix generation unit 42 generates the adjacency matrix A1 based on the set propagation threshold θ and the previously calculated violation propagation intensity matrix P. This process can be described by the following mathematical expression: , This binarization process can transform the continuous propagation intensity into a discrete network structure, which is convenient for subsequent network analysis.

[0085] Propagation Threshold is a numerical value used to screen out pairs of market players with significant communication potential. Specifically, only when the communication intensity between two market players is Greater than or equal to , they are considered to have a potential violation propagation relationship, and the corresponding adjacency matrix Elements in will be set to 1; otherwise, is set to 0.

[0086] Propagation Threshold The setting directly affects the adjacency matrix The sparsity of the network and the complexity of the network structure. By setting appropriate thresholds, the system can focus on the key nodes that are most likely to cause the spread of violations and avoid including all weak relationships in the analysis, thereby improving computational efficiency and accuracy of analysis. The binarized adjacency matrix makes the network structure clearer, which is convenient for further analysis using graph theory methods. A threshold that is too low will cause the network to be too dense and difficult to identify critical paths; a threshold that is too high may cause the network to be too sparse and miss important propagation relationships. Different road transport markets may have different competition intensities, transaction frequencies, and spatial proximity distributions. Therefore, the propagation threshold needs to be adjusted according to the specific market environment to ensure the applicability and robustness of the model.

[0087] Propagation Threshold The setting method uses the following steps:

[0088] First, calculate the transmission strength between all pairs of market entities , and draw its distribution graph (such as a histogram or density graph). According to the distribution of the propagation strength, select a suitable quantile as the threshold. For example, you can choose the 90th quantile (that is, 90% of the propagation strength is lower than this value) to ensure that only the strongest 10% of the relationships are retained. Finally, through cross-validation or other evaluation methods, verify whether the selected threshold can effectively distinguish strong and weak relationships. If the result is reasonable, use this quantile as the final propagation threshold Setting thresholds based on the statistical distribution of the data ensures that the threshold selection is objective and explainable, and is applicable to most market environments.

[0089] Or use clustering algorithms (such as K-means, DBSCAN, etc.) to measure the propagation intensity Perform cluster analysis to identify different levels of communication intensity. Based on the clustering results, select an appropriate threshold to distinguish strong relationships from weak relationships. For example, you can choose the boundary value between two major clusters as the threshold. Finally, verify whether the selected threshold can effectively distinguish strong relationships from weak relationships through cross-validation or other evaluation methods. Cluster analysis can automatically identify the natural stratification of communication intensity and is suitable for complex market environments, especially when the distribution of communication intensity is uneven.

[0090] In order to simplify the network structure, the system needs to set a propagation threshold ,Only when the transmission intensity exceeds this threshold, the two enterprises will be considered to have a potential violation transmission relationship.

[0091] Assume that the propagation threshold is set , then only when hour, ,otherwise For example, according to the previous calculation, ,therefore , indicating that there is a potential violation propagation relationship between A and B; and ,therefore , indicating that there is no obvious propagation relationship between A and C.

[0092] By setting appropriate thresholds, the system can focus on key nodes that are most likely to cause the spread of violations. For example, if the spread intensity between A and B is high, the system can prioritize the interaction between the two companies and take targeted regulatory measures.

[0093] The binary adjacency matrix makes the network structure clearer and facilitates further analysis using graph theory methods. For example, the system can identify which companies form a "violation transmission chain" through graph connectivity analysis.

[0094] Finally, the graph structure construction unit 43 constructs the final topological propagation network G based on the generated adjacency matrix A1. This network G can be regarded as a graph, in which nodes represent market entities and edges represent potential illegal behavior propagation relationships between entities. In this way, complex market relationships are transformed into intuitive network structures, laying the foundation for subsequent behavior evolution prediction.

[0095] Next, the system of the present invention introduces a behavior evolution prediction module 5. This module is connected to the topological network construction module 4, and its core task is to predict the evolution process of illegal behaviors based on the constructed topological propagation network G. The innovation of this module is that the traditional epidemiological model SIR (Susceptible-Infected-Recovered) model is improved and adapted to enable it to better describe the propagation process of illegal behaviors in the market.

[0096] The behavior evolution prediction module 5 contains three key sub-units. First, the state vector initialization unit 51 is responsible for initializing the violation state vector S(t) of the market entity. In this vector, each element represents the violation state of a market entity at a certain point in time, and its value range is [0, 1]. 0 represents full compliance, 1 represents full violation, and the intermediate value represents partial violation or risk state. This continuous state representation can better reflect the complex situation in reality than the traditional discrete state.

[0097] Secondly, the evolution parameter setting unit 52 is responsible for setting the propagation rate λ and the recovery rate μ. These two parameters largely determine the propagation speed of violations and the self-healing ability of the market. In practical applications, the values ​​of these parameters may need to be fitted and calibrated based on historical data. For example, in a market with strong supervision, the recovery rate μ may be higher; while in a market where information spreads rapidly, the propagation rate λ may be larger.

[0098] Finally, the state updating unit 53 calculates the updated value S(t+1) of the state vector based on the topological propagation network G, the state vector S(t), the propagation rate λ and the recovery rate μ. This updating process can be described by the following mathematical expression: , , in, is the adjacency matrix, Represents Hadamard product (element-wise multiplication). This model takes into account multiple factors such as network structure, current state, propagation rate, and recovery rate, and can more accurately simulate the propagation process of violations in the market.

[0099] The model is used to predict the spread of violations in the market. The state vector Indicates that each enterprise Violation status, transmission rate and recovery rate They represent the speed of transmission of violations and the self-healing ability of the market respectively.

[0100] Assume that at a certain point in time , the status of A is 0.6 (partial violation), the status of B is 0.3 (minor violation), and the status of C is 0.1 (almost compliant). According to the adjacency matrix , there is a propagation relationship between A and B, but not between A and C. If the propagation rate is The recovery rate is , then the state update at the next moment is as follows: A's status update: ; B's status update: ; The state of C remains unchanged because there is no propagation relationship.

[0101] By introducing the time dimension, the system can dynamically simulate the evolution of violations and provide predictions of future trends. For example, the system can predict whether certain companies will evolve from minor violations to serious violations, helping regulators intervene in advance. By adjusting different parameters, the system can perform multiple scenario analyses. For example, if the recovery rate is improved, , the system can evaluate the effectiveness of strengthening supervision; if the transmission rate is reduced , the impact of reducing the speed at which information spreads can be evaluated.

[0102] After the behavior evolution prediction, the system of the present invention introduces a risk assessment module 6. This module receives the output of the behavior evolution prediction module 5, performs risk assessment and generates early warning information. This step is crucial to realize the early warning function of the system.

[0103] The risk assessment module 6 includes three sub-units. First, the weight setting unit 61 is responsible for setting the weight coefficients of the current state, the change trend and the neighbor influence. The setting of these weights reflects the importance of different factors in risk assessment. For example, if the current state is considered to be the most important, it can be given a higher weight; if more attention is paid to the change trend, the corresponding weight can be increased.

[0104] Secondly, the risk calculation unit 62 calculates the risk value of each market entity based on the set weight coefficient and the prediction result of the illegal behavior evolution. This calculation process can be expressed by the following formula: , Among them, ω1, ω2, and ω3 are the weight coefficients of the current state, change trend, and neighbor influence, respectively. is the state value of subject i at time t, represents the rate of change of state, is an element in the adjacency matrix A1. This risk assessment model comprehensively considers the subject's own state, change trend and the impact of the surrounding environment, and can provide a more comprehensive risk assessment result.

[0105] The risk assessment model comprehensively considers the current status, changing trends and neighboring influences of market entities. and Used to adjust the importance of different factors.

[0106] Assume there are three logistics companies A, B, and C. At a certain point in time The status are , , 0.1. According to the adjacency matrix , there is a propagation relationship between A and B, but not between A and C. Assume that the weight coefficients are , , then the risk assessment of A is as follows: Current status: ; Change trend: Assume that the status of A has gradually deteriorated over the past period of time, and the change rate is ,but 0.05=0.015; Neighbor influence: A is influenced by B, Total risk value: ; Finally, the warning generation unit 63 generates warning information based on the calculated risk value and the preset warning threshold. For example, multiple risk levels can be set, and when the risk value exceeds a certain threshold, the system will issue a warning of the corresponding level. This hierarchical warning mechanism can help regulatory authorities take more targeted measures.

[0107] The system of the present invention also includes some auxiliary modules, which further improve the practicality and efficiency of the system. For example, the data acquisition module 1 can obtain data through multiple channels. Specifically, the module can collect network public opinion data from the Internet through web crawler technology, which includes user comments, news reports, etc. on social media. At the same time, it can also obtain regulatory data from government regulatory departments through API interfaces, such as corporate credit records, violation penalty information, etc. In addition, through the data exchange platform established with third-party payment companies, the system can obtain transaction data of market entities, which is crucial for analyzing market behavior patterns.

[0108] In order to ensure the quality and availability of data, the system of the present invention also includes a data preprocessing module 7. This module is responsible for cleaning, deduplication and formatting the collected raw data. It is particularly worth mentioning that for text data, the system also uses natural language processing technology for word segmentation and feature extraction. This step can convert unstructured text information into structured data that can be used for subsequent analysis.

[0109] Finally, in order to make the analysis results of the system more intuitive and easy to understand, the present invention also includes a visualization display module 8. This module can intuitively display the risk assessment results in the form of charts, heat maps, etc. More importantly, it supports interactive query and analysis, allowing users to explore data in depth and discover potential problems and trends.

[0110] Through the above detailed description, it can be seen how the road transport market violation situation awareness system of the present invention realizes the whole process from data collection, processing, analysis to final risk assessment and early warning through multiple closely cooperating modules. This systematic and intelligent method provides strong technical support for the supervision of the road transport market, and is expected to significantly improve the efficiency of supervision and promote the healthy development of the market. The road transport market violation situation awareness system of the present invention not only includes the aforementioned core functional modules, but also introduces a series of innovative method steps to further enhance the practicality and efficiency of the system. These method steps form a complete workflow, which enables the system to achieve comprehensive and accurate market violation situation awareness from data collection to final risk warning.

[0111] The model takes into account multiple factors and can more comprehensively assess the risk level of market entities. For example, A not only has a high level of violation status, but its status is also deteriorating. At the same time, it is affected by B, so its risk value is high. By introducing the trend of change, the system can capture the changes in the status of market entities in a timely manner and warn of potential risks in advance. For example, A's status is gradually deteriorating, and the system can remind the regulatory authorities to pay close attention. Considering the influence of neighbors, the system can identify market entities that are more affected by the surrounding environment. For example, A is more affected by B, and the system can suggest that the regulatory authorities strengthen supervision of B to prevent the further spread of violations.

[0112] The road transport market illegal behavior situation awareness system of the present invention realizes comprehensive awareness and early warning of illegal behaviors. Specifically:

[0113] Multi-dimensional data fusion: The system constructs the subject market relationship matrix R by comprehensively considering multi-dimensional data such as competition, transactions and geographical location, which can capture the complex relationships between market players more comprehensively.

[0114] Combination of history and reality: By introducing the historical violation matrix H, the system can evaluate the impact of historical violations on the current market and predict future propagation paths.

[0115] Dynamic simulation and scenario analysis: The improved SIR model enables the system to dynamically simulate the evolution of violations and supports multiple scenario analyses to help regulators develop optimal response strategies.

[0116] Comprehensive risk assessment: The risk assessment model comprehensively considers the current status, changing trends and neighboring impacts, and can provide more comprehensive risk assessment results to help regulators take timely measures.

[0117] These innovative applications not only improve the practicality and reliability of the system, but also provide strong technical support for the healthy and orderly development of the road transport market.

[0118] In a preferred embodiment of the present invention, the workflow of the system can be summarized into the following key steps: S1: Collect online public opinion, government supervision data and market entity transaction data, and perform pre-processing; S2: Based on the preprocessed data, construct the market entity relationship matrix R; S3: Based on the market subject relationship matrix R, calculate the violation propagation intensity matrix P; S4: Based on the violation propagation intensity matrix P, a topological propagation network G is constructed; S5: Predict the evolution of violation behaviors based on the topological propagation network G; S6: Based on the prediction results of the evolution of illegal behaviors, conduct risk assessment and generate early warning information.

[0119] This series of steps forms a logically rigorous, progressive analysis chain, and each step provides the necessary input for the next step, thus achieving comprehensive perception and prediction of market violations.

[0120] Specifically, in the data collection step (S1), the system not only collects online public opinion, government regulatory data, and market entity transaction data, but also preprocesses these data. This preprocessing process may include data cleaning, format unification, outlier processing, and other operations to ensure the data quality of subsequent analysis. For example, for text data, the system may use natural language processing technology to segment words and remove stop words; for numerical data, normalization may be performed to eliminate the impact of dimensions.

[0121] In the market subject relationship matrix construction step (S2), the system constructs the market subject relationship matrix R based on the preprocessed data. The construction process of this matrix involves complex mathematical operations and model design.

[0122] In the step (S3) of calculating the intensity of illegal behavior propagation, the system calculates the illegal behavior propagation intensity matrix P based on the market subject relationship matrix R. An innovation of the present invention is to introduce the historical illegal behavior matrix H and combine it with the relationship matrix R by means of Hadamard product. This calculation method not only considers the relationship between market subjects, but also considers the impact of historical illegal behaviors, so as to more accurately predict the propagation trend of illegal behaviors.

[0123] In the topological propagation network construction step (S4), the system constructs a topological propagation network G based on the violation propagation intensity matrix P. The key to this step is how to convert the continuous propagation intensity into a discrete network structure. The present invention adopts a threshold-based method, where θ is a preset propagation threshold. This method can effectively capture the most important propagation relationship while reducing the complexity of subsequent analysis.

[0124] In the violation behavior evolution prediction step (S5), the system predicts the evolution of the violation behavior based on the constructed topological propagation network G. The present invention innovatively adopts an improved SIR model.

[0125] Finally, in the risk assessment and warning information generation step (S6), the system performs risk assessment and generates warning information based on the prediction results of the evolution of illegal behaviors. The present invention proposes a risk assessment method that comprehensively considers the current state, change trend and neighbor influence. This multi-dimensional risk assessment method can more comprehensively characterize the risk status of market entities.

[0126] Through the above steps, the system of the present invention can realize comprehensive perception and prediction of illegal behaviors in the road transport market. This systematic and intelligent method not only improves the efficiency of supervision, but also provides a powerful support tool for decision makers, which helps to promote the healthy and orderly development of the road transport market.

[0127] It is worth noting that the system of the present invention has good scalability and adaptability. For example, in terms of data collection, the system can add new data sources or adjust data collection strategies as needed. In terms of model parameter setting, the system can automatically adjust various parameters through machine learning algorithms to adapt to different market environments and regulatory requirements. This flexibility enables the system to be continuously optimized and improved in practical applications, providing continuous technical support for long-term market supervision.

[0128] In order to verify the effectiveness and superiority of the road transport market illegal behavior situation awareness system of the present invention, a series of theoretical analysis and experimental verification are carried out. The following will introduce the settings, experimental methods and result analysis of the embodiments and comparative examples in detail.

[0129] Example 1: Complete system of the present invention

[0130] In Example 1, the complete system of the present invention is used, including all modules such as data collection, relationship matrix construction, propagation intensity calculation, topology network construction, behavior evolution prediction and risk assessment. The system parameters are set as follows: propagation threshold θ = 0.3; Propagation rate λ = 0.2; Recovery rate μ = 0.1; Risk assessment weights: ω1 = 0.5, ω2 = 0.3, ω3 = 0.2;

[0131] Comparative Example 1: Traditional Social Network Analysis Method

[0132] In Comparative Example 1, the traditional social network analysis method is used to predict market violations. This method mainly analyzes the direct relationship between market players and does not consider historical violations and complex communication dynamics.

[0133] Comparative Example 2: Simple Time Series Forecasting Method

[0134] In Comparative Example 2, simple time series forecasting methods, such as moving average and exponential smoothing, are used to independently predict the violations of each market entity without considering the mutual influence between the entities.

[0135] Experimental methods and evaluation indicators: The road transport market data of a province in the past three years was selected as the test data set. The data includes relevant information of 5,000 market entities, including transaction records, violation records, geographic locations, etc. The data of the first two years was used as the training set, and the data of the last year was used as the test set.

[0136] Evaluation metrics include:

[0137] 1. Prediction accuracy: The proportion of correctly predicted violations to the total predicted violations.

[0138] 2. Recall: The proportion of correctly predicted violations to actual violations.

[0139] 3. F1 score: The harmonic mean of prediction precision and recall.

[0140] 4. Average warning time: the average time interval from when the system issues a warning to when the actual violation occurs.

[0141] 5. False alarm rate: the ratio of false alarms to the total number of alarms.

[0142] The experimental results are as follows:

[0143] The following is the performance of the three methods on the test set:

[0144] The experimental results are analyzed as follows: It can be seen from the experimental results that the system of the present invention (Example 1) is significantly better than the traditional method in all evaluation indicators. The specific analysis is as follows:

[0145] 1. Prediction accuracy and recall rate: The prediction accuracy and recall rate of the system of the present invention reached 85.6% and 79.2% respectively, which is much higher than the other two methods. This shows that the system can more accurately identify potential violations and capture more actual violations. This advantage is mainly due to the system's comprehensive consideration of multiple factors such as market entity relationships, historical violations, and transmission dynamics.

[0146] 2. F1 score: The F1 score of the present invention is 0.823, which is more than 0.13 higher than the comparison method. This comprehensive index reflects that the system has achieved a good balance between precision and recall, and can provide more reliable decision support for regulatory authorities.

[0147] 3. Average warning time: The system of the present invention can issue warnings 14.5 days in advance, 6-8 days earlier than other methods. This significant time advantage provides sufficient time for regulatory authorities to take preventive measures and helps to intervene before violations occur.

[0148] 4. False alarm rate: The false alarm rate of the present invention is only 8.7%, which is much lower than other methods. This means that the system can more accurately identify real risks and reduce unnecessary waste of regulatory resources.

[0149] These experimental results fully demonstrate the superiority of the system of the present invention. Among them, the improvement of prediction accuracy and recall rate reflects the excellent ability of the system in identifying potential violations, which is mainly due to the multi-dimensional relationship matrix and improved SIR propagation model adopted by the system. The longer average warning time shows that the system can detect risks early, which is due to the system's topological network construction and behavior evolution prediction module. The lower false alarm rate reflects the accuracy of the system's risk assessment module, which is due to the comprehensive risk assessment method adopted by the system.

[0150] In general, the system of the present invention successfully improves the accuracy and timeliness of violation prediction and reduces the false alarm rate by integrating multi-source data, building complex network models, introducing propagation dynamics and realizing intelligent risk assessment. These advantages enable the system to provide more powerful technical support for the supervision of the road transport market, which is expected to significantly improve the supervision efficiency and promote the healthy and orderly development of the market.

[0151] In practical applications, the system of the present invention also shows good adaptability and scalability. For example, by adjusting parameters such as the propagation threshold θ, the propagation rate λ, and the recovery rate μ, the system can adapt to different types of market environments and violation behavior patterns. In addition, the weight setting of the risk assessment module can also be flexibly adjusted according to specific regulatory needs to highlight risk factors in different aspects.

[0152] Based on the above experimental results and analysis, it is believed that Example 1 represents the best implementation of the present invention. In this example, the system gives full play to the advantages of each module and achieves the goals of high precision, low false alarms, and early warning. In particular, when the propagation threshold θ is set to 0.3, the propagation rate λ is 0.2, and the recovery rate μ is 0.1, the system shows the best balance. This set of parameter settings can capture important propagation relationships without causing the network to be too dense, and also reflects the propagation and recovery characteristics of violations in reality.

[0153] In addition, the setting of risk assessment weights (ω1 = 0.5, ω2 = 0.3, ω3 = 0.2) has also been proven to be very effective. This weight distribution reflects the relative importance of the current status, change trend and neighbor influence in risk assessment, and can comprehensively and accurately characterize the risk status of market entities.

[0154] In conclusion, the road transport market illegal behavior situation awareness system of the present invention has shown significant performance advantages through innovative technical solutions and reasonable parameter settings. This not only verifies the effectiveness of the system, but also lays a solid foundation for further optimization and promotion and application in the future.

[0155] It should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. The road transport market illegal behavior situation awareness system is characterized by: include: Data acquisition module for: Collect online public opinion, government regulatory data, and market entity transaction data; Preprocess and store the collected data; A relationship matrix building module is connected in communication with the data acquisition module and is used to: Receiving the preprocessed data sent by the data acquisition module; Based on the preprocessed data, a market subject relationship matrix R is constructed; The propagation intensity calculation module is communicatively connected with the relationship matrix construction module and is used for: Receive the market entity relationship matrix R sent by the relationship matrix building module; Based on the market subject relationship matrix R, the violation propagation intensity matrix P is calculated; A topology network construction module is communicatively connected with the propagation intensity calculation module and is used for: Receiving the violation propagation intensity matrix P sent by the propagation intensity calculation module; Based on the violation propagation intensity matrix P, a topological propagation network G is constructed; The behavior evolution prediction module is connected to the topology network construction module for: Receiving the topology propagation network G sent by the topology network construction module; Based on the topological propagation network G, predict the evolution of illegal behaviors; A risk assessment module is connected in communication with the behavior evolution prediction module and is used to: Receiving the illegal behavior evolution prediction result sent by the behavior evolution prediction module; Based on the predicted results of the evolution of the illegal behavior, risk assessment is performed and early warning information is generated.

2. The system according to claim 1, characterized in that The relationship matrix building module includes: A relationship factor extraction unit, used to extract a competitive relationship index, a transaction frequency index and a spatial proximity index from the preprocessed data; A weight calculation unit, used to calculate the weight coefficients of the competition relationship index, the transaction frequency index and the spatial proximity index; The matrix generating unit is used to generate a market subject relationship matrix R based on the index extracted by the relationship factor extracting unit and the weight coefficient calculated by the weight calculating unit.

3. The system according to claim 1, characterized in that The propagation intensity calculation module comprises: A historical violation matrix generating unit, used for generating a historical violation matrix H based on the preprocessed data; The Hadamard product calculation unit is used to calculate the Hadamard product of the market entity relationship matrix R and the historical violation matrix H to obtain the violation propagation intensity matrix P.

4. The system according to claim 1, characterized in that The topology network building module includes: A threshold setting unit, used for setting a propagation threshold θ; An adjacency matrix generating unit, used for generating an adjacency matrix A1 based on the propagation threshold θ and the violation propagation intensity matrix P; A graph structure building unit is used to build a topological propagation network G based on the adjacency matrix A1.

5. The system according to claim 1, characterized in that The behavior evolution prediction module includes: A state vector initialization unit, used to initialize the violation state vector S(t) of the market entity; An evolution parameter setting unit, used to set the propagation rate λ and the recovery rate μ; The state updating unit is used to calculate the updated value S(t+1) of the state vector based on the topological propagation network G, the state vector S(t), the propagation rate λ and the recovery rate μ.

6. The system according to claim 1, characterized in that The risk assessment module includes: A weight setting unit is used to set the weight coefficient of the current state, change trend and neighbor influence; A risk calculation unit, used to calculate the risk value of each market entity based on the weight coefficient and the prediction result of the evolution of the illegal behavior; The warning generation unit is used to generate warning information according to the calculated risk value and the preset warning threshold.

7. The system according to claim 1, characterized in that The data collection module collects online public opinion data from the Internet through web crawler technology, obtains regulatory data from government regulatory departments through API interfaces, and obtains market entity transaction data from third-party payment companies through data exchange platforms.

8. The system according to claim 1, characterized in that It also includes a data preprocessing module, which is used to clean, deduplicate, and unify the format of the collected raw data, and use natural language processing technology to segment and extract features from text data.

9. The system according to claim 1, characterized in that It also includes a visual display module, which is used to intuitively display risk assessment results in the form of charts and heat maps, and supports interactive query and analysis.

10. A method for sensing illegal behaviors in the road transport market based on the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1: Collect online public opinion, government supervision data and market entity transaction data, and perform pre-processing; S2: Based on the preprocessed data, construct the market entity relationship matrix R; S3: Based on the market subject relationship matrix R, calculate the violation propagation intensity matrix P; S4: Based on the violation propagation intensity matrix P, a topological propagation network G is constructed; S5: Predict the evolution of violation behaviors based on the topological propagation network G; S6: Based on the prediction results of the evolution of illegal behaviors, conduct risk assessment and generate early warning information.

Images