Risk early warning platform and early warning method for rural property right transaction

Abstract

The invention provides a risk early warning platform and early warning method for rural property trading, and relates to the technical field of trading risk early warning, and the platform comprises a data collection module which is used for obtaining a basic data set; the similar aggregation module is used for determining an aggregated basic data cluster; the frequent item mining module is used for determining a risk transaction item set and a risk transaction item basic data cluster; the network training module is used for constructing an adversarial risk identification network layer; the risk identification module is used for determining an adversarial risk identification result set; and the early warning information generation module is used for generating risk early warning information. The technical problems of inaccurate transaction risk identification and low risk early warning efficiency caused by factors such as information asymmetry and transaction mode diversification in the rural property right transaction process are solved, the risk items are identified through the adversarial risk identification network layer, the transaction risk identification accuracy and the risk early warning efficiency are improved, and the risk early warning efficiency is improved. And the prevention and control capability of the rural property right transaction risk is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of transaction risk early warning, and in particular to a risk early warning platform and early warning method for rural property rights transactions. Background Art

[0002] In the process of rural property rights transactions, with the development of information technology and the application of transaction platforms, more and more rural property rights transaction projects are conducted in a digital way. Rural property rights transactions involve important assets such as land use rights and real estate, and the transaction methods include various forms such as assignment, transfer, contracting, and leasing. Different transaction methods will lead to different transaction risks, which may affect the fairness and security of transactions and even affect the healthy development of the rural property rights market. The data security risks in rural property rights transactions are particularly prominent, mainly including information leakage and data tampering. The transaction process involves a large amount of sensitive information of individuals or enterprises. If this information is leaked, it may seriously affect the privacy security of transaction participants; secondly, if the transaction records and contract content are maliciously tampered with, it will affect the integrity and authenticity of the data and pose a threat to the legitimacy and fairness of the transaction. In addition, the differences in transaction conditions between similar transaction projects in different transaction projects in the same period of time, and the differences between them and historical similar transaction projects, can often reveal potential transaction risks. For example, if a certain type of project has an abnormal transaction frequency or price fluctuation, it will imply the risk factors of such projects. Summary of the invention

[0003] This application provides a risk warning platform and warning method for rural property rights transactions, aiming to solve the technical problems of inaccurate transaction risk identification and low risk warning efficiency caused by factors such as information asymmetry and diversified transaction methods in the process of rural property rights transactions.

[0004] The first aspect disclosed in the present application provides a risk warning platform for rural property rights transactions, the platform comprising: a data collection module, used to perform basic data collection of items to be traded, and obtain K basic data sets of K items to be traded, wherein the items to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1; a homogeneous aggregation module, used to perform homogeneous aggregation on the K basic data sets with the transaction method as the clustering target, and determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction method identifiers, and M is an integer less than or equal to K; a frequent item mining module, used to interact with the M transaction method identifiers to perform risk transactions Frequent item mining, determining M risky transaction project sets and M risky transaction project basic data clusters; a network training module, used to obtain M interference transaction project basic data clusters based on the M risky transaction project sets, and train the adversarial risk identification network layer in combination with the M risky transaction project basic data clusters to construct M adversarial risk identification network layers; a risk identification module, used to use the M adversarial risk identification network layers to identify transaction risks for the M aggregated basic data clusters, and determine M adversarial risk identification result sets; a warning information generation module, used to generate risk warning information when there are risky items to be traded in the M adversarial risk identification result sets.

[0005] The second aspect disclosed in the present application provides a risk warning method for rural property rights transactions, which is implemented through the above-mentioned risk warning platform for rural property rights transactions, and the method includes: executing basic data collection of projects to be traded to obtain K basic data sets of K projects to be traded, wherein the projects to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1; taking the transaction method as the clustering target, performing similar aggregation on the K basic data sets to determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction method identifiers, and M is an integer less than or equal to K; interacting with the M transaction mode identifiers are used to mine frequent risk transaction items, and M risk transaction project sets and M risk transaction project basic data clusters are determined; based on the M risk transaction project sets, M interference transaction project basic data clusters are obtained, and the adversarial risk identification network layer is trained in combination with the M risk transaction project basic data clusters to construct M adversarial risk identification network layers; the M adversarial risk identification network layers are used to identify transaction risks for the M aggregated basic data clusters to determine M adversarial risk identification result sets; when there are risky items to be traded in the M adversarial risk identification result sets, risk warning information is generated.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The method first collects the basic data of the project to be traded to form a basic data set of K projects, so that each project corresponds to its basic data one by one. Then, these data are aggregated with the transaction method as the clustering target to generate M aggregated basic data clusters, each cluster corresponding to a transaction method, so as to achieve effective grouping and organization of data. This aggregation process helps to refine and classify the data based on the transaction method, making the subsequent risk analysis more targeted. Subsequently, the frequent items of risky transactions are mined by interacting with M transaction method identifiers to identify the project set with transaction risks and the corresponding basic data clusters. This mining step can discover potential anomalies and high-frequency risk items in the transaction mode, making the characteristics of high-risk projects clearer and further improving the accuracy of risk identification. After that, the interference transaction data clusters are obtained from the risk project set, and these interference data are combined with the risk transaction data. Through the training of the adversarial risk identification network layer, M adversarial risk identification network layers are gradually constructed. The training process of the network layer enhances the model's sensitivity to risk and anti-interference ability by simulating risk interference, so that the model can more accurately distinguish abnormal transaction patterns when identifying risks. Finally, the trained adversarial risk identification network layer is used to conduct comprehensive risk identification on the aggregated basic data cluster, generating M risk identification result sets. When a risky transaction project is detected, risk warning information is automatically generated. Through this series of operations, not only can transaction risks be discovered in advance, but also the accuracy and timeliness of risk identification can be effectively improved, ensuring the security of rural property rights transactions.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic diagram of the structure of a risk warning platform for rural property rights transactions provided in an embodiment of the present application.

[0010] Figure 2 A flow chart of a risk warning method for rural property rights transactions provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: data collection module 1, similar aggregation module 2, frequent item mining module 3, network training module 4, risk identification module 5, early warning information generation module 6. DETAILED DESCRIPTION

[0012] The embodiments of the present application solve the technical problems of inaccurate transaction risk identification and low risk warning efficiency caused by factors such as information asymmetry and diversified transaction methods in the process of rural property rights transactions by providing a risk warning platform and warning method for rural property rights transactions.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a risk warning platform for rural property rights transactions, and the platform includes:

[0015] The data collection module 1 is used to collect basic data of the items to be traded, and obtain K basic data sets of K items to be traded, wherein the items to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1.

[0016] Specifically, in the data collection module 1, for each project to be traded, basic information directly related to the project is obtained and recorded, such as project type, location, asset attributes, transaction amount, transaction period, historical transaction records, etc., to ensure that the collected basic data can fully cover the key elements of each project. In this way, each project to be traded will have a unique corresponding basic data set, so that classification and analysis can be carried out based on these data. This process not only provides detailed data support for subsequent risk identification, but also establishes a standardized data file for each project to ensure the integrity and consistency of the data, laying a solid foundation for subsequent aggregation and identification links.

[0017] The same type aggregation module 2 is used to perform the same type aggregation on the K basic data sets with the transaction mode as the clustering target, and determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction mode identifiers, and M is an integer less than or equal to K.

[0018] Specifically, in the same type aggregation module 2, the basic data set of each project to be traded is analyzed to extract the transaction method information of each project. The transaction methods include various forms such as transfer, assignment, contracting, leasing, etc. Subsequently, the projects with the same transaction method and the corresponding basic data set are classified into the same data cluster. For example, all projects with transfer methods are aggregated into one data cluster, and the projects with transfer methods are aggregated into another data cluster, and so on. In this process, a corresponding transaction method identifier is also assigned to each aggregated basic data cluster to ensure that each data cluster can distinguish its transaction method through a unique identifier. For example, the aggregated basic data cluster of the transfer method is marked as 0001, and the aggregated basic data cluster of the transfer method is marked as 0010, so that it can be quickly called and identified in subsequent analysis. Through the above clustering, K basic data sets are gathered together to form M aggregated basic data clusters, and each aggregated basic data cluster corresponds to a transaction method identifier. Among them, the number of M is usually less than or equal to K. After clustering is completed, the system terminal can effectively organize the data, making it easier to analyze and identify risk patterns specific to different transaction methods in subsequent steps, thereby improving the accuracy and efficiency of risk analysis.

[0019] The frequent item mining module 3 is used to interact with the M transaction mode identifiers to perform risk transaction frequent item mining, and determine M risk transaction item sets and M risk transaction item basic data clusters.

[0020] Specifically, in the frequent item mining module 3, M transaction mode identifiers corresponding to each other are obtained by interacting with M aggregated data clusters, and then the relevant data of risky transaction projects are collected based on these transaction mode identifiers. Subsequently, by performing multi-point aggregation analysis on the collected data, the aggregation characteristics of different risk factors are extracted, and the aggregation group set of each factor is determined. Then, according to the aggregation group set of each factor, the collected data is subjected to strategy extraction to screen out a set of high-risk transaction projects. These sets contain projects that are often traded under high-risk conditions, which is convenient for further risk monitoring. After that, the risky transaction project set is matched one-to-one with the basic data cluster to generate a basic data cluster of risky transaction projects, laying the foundation for subsequent risk identification and early warning. Through the above steps, high-risk transaction patterns and projects can be identified and extracted, the sensitivity to potential risky transactions can be improved, and clear risk items can be provided for further risk identification and early warning.

[0021] Furthermore, the platform comprises:

[0022] Based on big data, relevant data of risk transaction projects are collected according to the M transaction mode identifiers to obtain M initial risk transaction project sets, M initial risk transaction project basic data clusters and M initial risk transaction project risk factor sets; based on the M initial risk transaction project factor sets, multi-point factor aggregation analysis is performed to determine the M initial risk transaction project factor cluster group sets; according to the factor quantity in each initial risk transaction project factor cluster group in the M initial risk transaction project factor cluster group sets, strategy extraction is performed on the M initial risk transaction project sets to determine the M risk transaction project sets; based on a one-to-one mapping relationship between risk transaction projects and risk transaction project basic data, the M initial risk transaction project basic data clusters are mapped and matched according to the M risk transaction project sets to obtain the M risk transaction project basic data clusters.

[0023] Preferably, for each transaction method identifier, by performing big data analysis on various online transaction platforms, third-party credit reporting agencies and other data sources, the transaction records and historical data related to each transaction method are screened out, including transaction frequency, amount, participants, historical risk records, etc., and the projects with potential risks in each transaction method are determined, and these projects are classified and stored to form M initial risk transaction project sets. For each initial risk transaction project set, the corresponding basic data and risk factors are respectively integrated to form M initial risk transaction project basic data clusters and M initial risk transaction project risk factor sets, so as to uniformly analyze and manage risk projects. Subsequently, the initial risk transaction project factor set is subjected to multi-point aggregation analysis to identify and aggregate factors with similar risk characteristics, and generate M initial risk transaction project factor cluster sets. This aggregation process is carried out through the similarity analysis of the initial risk transaction project factors, which can effectively summarize the aggregation characteristics of the risk factors and help better identify similar risk patterns. In the initial risk transaction project factor cluster set, the initial risk transaction project set is subjected to strategy extraction according to the number of factors in each factor cluster and the preset total amount of extraction. This step determines the final set of risky transaction projects and screens out the most risky representative projects to ensure that subsequent analysis is more accurate. Afterwards, based on the one-to-one mapping relationship between risky transaction projects and basic data, the screened risky transaction project set is matched with the initial risky transaction project basic data cluster to obtain the final risky transaction project basic data cluster. This mapping enables each risky project to have a unique data cluster association, ensuring the integrity and accuracy of subsequent risk identification. Through the above-mentioned multi-layer screening and precise mapping of high-risk transaction projects, the accuracy of identifying high-risk transaction projects is improved, and the data foundation and analytical support for subsequent risk identification and early warning are laid.

[0024] Furthermore, based on the M initial risk transaction project factor sets, a factor multi-point aggregation analysis is performed to determine the M initial risk transaction project factor aggregation group sets, including:

[0025] Randomly extract M extracted initial risk transaction project factor sets from the M initial risk transaction project factor sets as M initial clustering factor sets; traverse the M initial clustering factor sets respectively to perform pairwise initial clustering factor similarity analysis to determine M similarity sets; determine whether the number of similarities in the M similarity sets whose similarities exceed a preset similarity is greater than or equal to a preset number, and if so, randomly extract M extracted initial risk transaction project factor sets from the M initial risk transaction project factor sets again to update the M initial clustering factor sets.

[0026] Optionally, in the M initial risk transaction project factor sets, several factors are randomly extracted from each factor set to form M extracted initial risk transaction project factor sets as M initial clustering factor sets. The purpose of random extraction is to generate initial data points for subsequent similarity analysis, ensure the diversity and randomness of the data, and help avoid overfitting. Then, for example, for factors A1, A2, A3 in an initial clustering factor set A, all possible factor pair combinations are generated, such as (A1, A2), (A1, A3), (A2, A3). For each pair of factors (such as A1 and A2), the Euclidean distance is used as a measure of similarity to calculate the similarity distance between each pair of factors. The smaller the similarity distance, the greater the similarity between the two factors. After that, the calculated similarity distance is added to 1, and the ratio of 1 to the sum is calculated to obtain the similarity between each pair of factors. After obtaining the similarity between each pair of factors, the similarity of each pair of factors is recorded in the corresponding similarity set, and each similarity set corresponds to an initial clustering factor set. Repeat the above process to generate M similarity sets corresponding to the M initial clustering factor sets to measure the internal similarity of the factor sets. After obtaining the M similarity sets, determine the number of similarities in each set that exceed the preset similarity. If the number of similarities that exceed the preset similarity reaches or exceeds the preset number, it means that the factors in these initial clustering factor sets have high similarity and cannot fully represent the diverse risk characteristics. At this time, new factors will be randomly extracted from the initial risk transaction project factor set to replace the existing initial clustering factor set, thereby generating new M clustering factor sets. In this way, multiple extractions and comparisons can ensure the diversity of data in the factor set, making the subsequent clustering analysis more representative.

[0027] Furthermore, the platform comprises:

[0028] If not, then taking the M initial clustering factor sets as clustering points respectively, a multi-point clustering analysis is performed on the M initial risk transaction project factor sets to determine the M initial risk transaction project factor clustering group sets.

[0029] Optionally, when the number of similarities in the M similarity sets that exceeds the preset similarity is less than the preset number, M initial clustering factors are randomly selected from the screened M initial clustering factor sets without replacement as clustering points. These points will become the core of clustering analysis, which is used to judge the similarity and belonging of other factors to these points. Subsequently, for each factor in the initial risk transaction project factor set, the distance between it and each clustering point is calculated one by one, and each factor is allocated in combination with the preset clustering step length to obtain M initial risk transaction project factor sets. After the factor allocation is completed, M initial clustering factors are randomly selected from the M initial clustering factor sets without replacement again to perform clustering analysis on the M initial risk transaction project factor sets, and determine M initial risk transaction project factor clustering group sets. These clustering group sets reflect the similar risk characteristics around each clustering point, which will be used to identify project groups with similar risk factors, and provide a classification basis for further risk detection.

[0030] Furthermore, the platform comprises:

[0031] M first initial clustering factors are randomly extracted from the M initial clustering factor sets as M first clustering points without replacement; M first initial risk transaction project factor clustering groups are constructed based on the M initial clustering factor sets with the M first clustering points as the centers of the M first clustering groups and a preset clustering step as the radius; the M first initial risk transaction project factor clustering groups are diffused outward from the edges according to the preset clustering step to obtain M first diffused initial risk transaction project factor clustering groups.

[0032] Optionally, a factor is randomly selected from each initial clustering factor set without replacement as M first clustering points, and these points will be used as the initial clustering centers. Subsequently, with each first clustering point as the center and the preset clustering step length as the radius, the factors within the step length range from the first clustering point are searched in the initial clustering factor set by Euclidean distance, and these factors are divided into the corresponding initial risk transaction project factor clustering groups. This step ensures that the factors around the center of each clustering group belong to the same clustering group, thereby forming M first initial risk transaction project factor clustering groups. After the initial clustering group is constructed, the edge factors of each clustering group are diffused outward according to the preset clustering step length, that is, at the edge point of each clustering group, the factors within the step length range are continued to be searched by Euclidean distance. Factors that meet the step length range are added to the clustering group, and the diffusion operation is repeated until no more qualified factors can be found, thereby forming M first diffused initial risk transaction project factor clustering groups, ensuring that the clustering group contains as many similar factors as possible, laying the foundation for the accurate aggregation of risk factors.

[0033] Calculating the M first clustering densities of the M first initial risk transaction project factor clustering groups and the M first diffusion clustering densities of the M first diffusion initial risk transaction project factor clustering groups; when the M first diffusion clustering densities are less than the M first clustering densities, stopping the diffusion update, and removing the M first initial risk transaction project factor clustering groups from the M initial risk transaction project factor sets, to obtain an updated M initial risk transaction project factor sets;

[0034] Optionally, in each diffusion, for each first initial risk transaction project factor cluster group, the ratio of the number of factors in the first initial risk transaction project factor cluster group to the area of ​​the first initial risk transaction project factor cluster group is calculated to obtain the first clustering density. The area of ​​the first initial risk transaction project factor cluster group is calculated based on the circular area formula and the preset clustering step. For each first diffusion initial risk transaction project factor cluster group, the ratio of the number of factors in the first diffusion initial risk transaction project factor cluster group to the area of ​​the first diffusion initial risk transaction project factor cluster group is calculated to obtain the first diffusion clustering density. The area of ​​the first diffusion initial risk transaction project factor cluster group is calculated based on the circular area formula and 2 times the preset clustering step. Repeat the above process to calculate M first clustering densities of M first initial risk transaction project factor cluster groups and M first diffusion clustering densities of M first diffusion initial risk transaction project factor cluster groups. Subsequently, for each diffusion cluster group, the first diffusion clustering density is compared with the first clustering density. If the first diffusion aggregation density is less than the first aggregation density, it indicates that the density decreases during the factor diffusion process and has entered the low-density area, which is no longer suitable for further expansion. At this time, the diffusion update operation of the cluster group will be stopped. When all cluster groups stop diffusing, the corresponding M first initial risk transaction project factor cluster groups are removed from the initial risk transaction project factor set. This operation clears the factors that have been completely aggregated, ensuring that the subsequent operations only process factors that have not yet been aggregated. After the above steps, the updated initial risk transaction project factor set after eliminating the cluster group is obtained. This set only retains the factors that have not been completely aggregated, providing a data basis for subsequent aggregation and diffusion operations.

[0035] Again, M second initial clustering factors are randomly extracted from the M initial clustering factor sets as M second clustering points without replacement, and clustering analysis is performed on the updated M initial risk transaction project factor sets to obtain M second initial risk transaction project factor clustering groups; based on the M initial clustering factor sets, multi-point clustering analysis is performed on the M initial risk transaction project factor sets to obtain the M initial risk transaction project factor clustering group sets.

[0036] Optionally, after the update of the M initial risk transaction project factor sets is completed, M factors are randomly extracted from the initial aggregation factor set again without replacement as the second initial aggregation factors, and they are used as M second aggregation points. This step obtains a new aggregation center from the screened initial factor set, providing a starting point for the subsequent update and improvement of the aggregation group. Subsequently, the M extracted second aggregation points are used as the center to perform aggregation analysis on the updated initial risk transaction project factor set. At this time, the distance between each factor and the second aggregation point is calculated in the same way as above, and the factors whose distance from the aggregation point meets the preset aggregation step length are classified into the corresponding second initial risk transaction project factor aggregation group. This step ensures that the factors are reasonably aggregated according to the latest aggregation point characteristics, and gradually form M more stable second initial risk transaction project factor aggregation groups. After the M second initial risk transaction project factor aggregation groups are initially formed, the same diffusion process as above is performed, and the M second initial risk transaction project factor aggregation groups are eliminated from the M initial risk transaction project factor set to complete the update of the M initial risk transaction project factor set. Afterwards, the above process is repeated based on the M initial aggregation factor sets and the updated M initial risk transaction project factor sets until all factors in the M initial risk transaction project factor sets are aggregated. Then, all generated initial risk transaction project factor aggregation groups are aggregated, that is, the M first initial risk transaction project factor aggregation groups, the M second initial risk transaction project factor aggregation groups, etc. are aggregated to obtain M initial risk transaction project factor aggregation group sets, which provide detailed factor aggregation results for subsequent risk identification and analysis, so that risk detection has stronger identification accuracy and comprehensiveness.

[0037] Further, according to the factor amount in each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets, mapping is performed on the M initial risk transaction project sets to extract strategies, and determining the M risk transaction project sets, including:

[0038] The number of factors of each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets is counted respectively to obtain M factor number sets; each factor number in the M factor number sets is divided by the sum of the factor numbers in the corresponding M factor number sets to obtain M extraction coefficient sets; a preset total extraction amount is obtained, and strategy extraction is performed on the M initial risk transaction project sets based on the M extraction coefficient sets and the preset total extraction amount to determine the M risk transaction project sets.

[0039] Optionally, after obtaining M initial risk transaction project factor clustering groups, for each initial risk transaction project factor clustering group set, the number of factors contained in each initial risk transaction project factor clustering group is counted, thereby obtaining M factor number sets. Subsequently, the number of each factor in the M factor number sets is divided by the sum of the number of all factors in the factor number set to which it belongs, to obtain M extraction coefficients, which indicate the relative weight of each clustering group in the whole, and are used to guide the subsequent extraction ratio. After that, a preset total extraction amount is obtained, which represents the total number of factors to be extracted from all initial risk transaction project sets, and then each extraction coefficient is multiplied by the total extraction amount to calculate the number of factors to be extracted from each clustering group, thereby forming a specific extraction strategy. Then, according to the extraction amount of each clustering group in the extraction strategy, a corresponding number of projects are extracted from the corresponding M initial risk transaction project sets to form M risk transaction project sets. The number of projects extracted from each set meets the extraction ratio of its clustering group, ensuring that the characteristics of different clustering groups can be distributed in the final risk transaction project set according to the weight, improving the comprehensiveness and accuracy of risk identification, and providing a balanced and representative risk factor set for subsequent analysis.

[0040] The network training module 4 is used to obtain M interference transaction project basic data clusters based on the M risk transaction project sets, train the adversarial risk identification network layer in combination with the M risk transaction project basic data clusters, and construct M adversarial risk identification network layers.

[0041] Specifically, based on M risky transaction project sets, M interference transaction project basic data clusters are generated. These interference clusters are composed of ordinary transaction data similar to risky transaction projects but not marked as high-risk, and are used to construct interference samples in the adversarial risk identification network layer, so that the network layer can more accurately identify the real risk characteristics. Subsequently, the M risky transaction project sets, the M risky transaction project basic data clusters and the M interference transaction project basic data clusters are paired and combined to form M groups of training data sets. Each training data set contains positive samples (risky project data) and negative samples (interference project data), providing a data basis for the subsequent training of the adversarial network. Afterwards, by alternately training the generator and the discriminator in the adversarial network, the generator and the discriminator are mutually opposed and jointly optimized until the generator and the discriminator reach a balanced state, thereby constructing M independent adversarial risk identification network layers, each of which is optimized for a specific risky transaction project set and the corresponding interference data. These network layers can more accurately identify transaction projects with specific risk patterns, thereby enhancing the accuracy in identifying high-risk transaction projects and providing stronger reliability for subsequent risk warnings.

[0042] Furthermore, the platform comprises:

[0043] Construct an adversarial risk identification network layer, wherein the adversarial risk identification network layer includes a generator and a discriminator; respectively use the M interference transaction project basic data clusters, the M risk transaction project sets and the M risk transaction project basic data clusters to train the generator and the discriminator until convergence, thereby obtaining the trained M adversarial risk identification network layers.

[0044] Preferably, based on the adversarial network, a confrontation risk identification network layer is constructed, and the confrontation risk identification network layer includes a generator and a discriminator. The generator is responsible for generating forged data similar to real risk trading projects to deceive the discriminator, while the task of the discriminator is to distinguish between real risk data and fake data generated by the generator. This structure allows the generator and the discriminator to confront each other and continuously improve each other's recognition and forgery capabilities. After constructing the structure of the confrontation risk identification network layer, M interference trading project basic data clusters, M risk trading project sets, and M risk trading project basic data clusters are respectively input into the generator and the discriminator to establish labels for real and forged data. Among them, the risk trading project data and the risk trading project basic data cluster are input into the discriminator as real labels, and the forged data and interference data clusters generated by the generator are input as pseudo labels. This dual data input allows the generator to continuously generate realistic data while also allowing the discriminator to improve its sensitivity to real risk features. During the training process, the generator and the discriminator are optimized alternately. The discriminator fixes the parameters of the generator and then updates itself. By comparing real and forged data, the discriminator gradually learns to identify the characteristics of forged data. On the contrary, when the parameters of the discriminator are fixed, the generator updates its own parameters to generate fake data that is closer to the real thing, trying to deceive the discriminator so that it cannot easily distinguish. This process continuously adjusts the network parameters by calculating the cross-entropy loss of the generator and the discriminator. When the generator and the discriminator reach a balance after multiple iterations, that is, the fake data generated by the generator is realistic enough, and the classification accuracy of the discriminator can no longer be significantly improved, the training reaches a convergence state. The M adversarial risk identification network layers finally formed have extremely strong discrimination capabilities, which can effectively distinguish between high-risk transaction project data and forged data, thereby realizing efficient identification of real risk transaction projects and providing solid support for subsequent risk warnings.

[0045] The risk identification module 5 is used to use the M counter-risk identification network layers to perform transaction risk identification on the M aggregated basic data clusters and determine M counter-risk identification result sets.

[0046] Specifically, in the risk identification module 5, risk identification is performed on M aggregated basic data clusters through the trained M adversarial risk identification network layers. Each adversarial network layer processes one aggregated basic data cluster respectively, and compares the data in the cluster with the risk characteristics identified by the network layer to identify the potential risk transactions therein. The identification results will be summarized into M adversarial risk identification result sets, including the transaction project information determined to be risky in each aggregated basic data cluster. Through this process, potential risk projects in different transaction clusters can be accurately screened out, forming clear risk results, and providing data support for the subsequent generation of effective early warning information.

[0047] The warning information generating module 6 is used to generate risk warning information when there are risky items to be traded in the M counter-risk identification result sets.

[0048] Specifically, in the early warning information generation module 6, the contents of the generated M sets of counter-risk identification results are checked to confirm whether there are any pending transaction items marked as risks. Once any potential risk items are found in the identification results, corresponding risk early warning information will be generated immediately. The risk early warning information will include the specific characteristics of the risk items, the transaction methods they belong to, etc., so that relevant users can understand and pay attention to these items in a timely manner. This early warning mechanism ensures that risky transactions can be quickly noticed, helping relevant users to take necessary precautions before transactions occur to ensure transaction security.

[0049] In summary, the risk warning platform for rural property rights transactions provided by the embodiments of the present application has the following technical effects:

[0050] The data collection module 1 is used to collect basic data of the items to be traded, and obtain K basic data sets of K items to be traded, wherein the items to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1; the same type aggregation module 2 is used to perform the same type aggregation on the K basic data sets with the transaction mode as the clustering target, and determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction mode identifiers, and M is an integer less than or equal to K; the frequent item mining module 3 is used to perform risk transaction frequent item mining by interacting with the M transaction mode identifiers, and determine M risk transaction project sets and M risky transaction project basic data clusters; network training module 4, used to obtain M interference transaction project basic data clusters based on the M risky transaction project sets, and train the confrontation risk identification network layer in combination with the M risky transaction project basic data clusters to construct M confrontation risk identification network layers; risk identification module 5, used to use the M confrontation risk identification network layers to identify transaction risks for the M aggregated basic data clusters, and determine M confrontation risk identification result sets; warning information generation module 6, used to generate risk warning information when there are risky transaction projects in the M confrontation risk identification result sets. Through the above steps, the technical problems of inaccurate transaction risk identification and low risk warning efficiency caused by factors such as information asymmetry and diversified transaction methods in the process of rural property rights transactions are solved, and the effect of identifying risk projects through the confrontation risk identification network layer, improving the accuracy of transaction risk identification and risk warning efficiency, and enhancing the ability to prevent and control rural property rights transaction risks is achieved.

[0051] Embodiment 2, based on the same inventive concept as the risk warning platform for rural property rights transactions in the above embodiments, Figure 2 As shown, the embodiment of the present application provides a risk warning method for rural property rights transactions, the method comprising:

[0052] Perform basic data collection of projects to be traded to obtain K basic data sets of K projects to be traded, wherein the projects to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1; take the transaction method as the clustering target, perform the same type aggregation on the K basic data sets, and determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction method identifiers, and M is an integer less than or equal to K; perform risk transaction frequent item mining by interacting with the M transaction method identifiers, and determine M risk transaction project sets and M risk transaction project basic data clusters; obtain M interference transaction project basic data clusters based on the M risk transaction project sets, and train the adversarial risk identification network layer in combination with the M risk transaction project basic data clusters to construct M adversarial risk identification network layers; use the M adversarial risk identification network layers to perform transaction risk identification on the M aggregated basic data clusters, and determine M adversarial risk identification result sets; when there are risky projects to be traded in the M adversarial risk identification result sets, generate risk warning information.

[0053] Furthermore, the method comprises:

[0054] Based on big data, relevant data of risk transaction projects are collected according to the M transaction mode identifiers to obtain M initial risk transaction project sets, M initial risk transaction project basic data clusters and M initial risk transaction project risk factor sets; based on the M initial risk transaction project factor sets, multi-point factor aggregation analysis is performed to determine the M initial risk transaction project factor cluster group sets; according to the factor quantity in each initial risk transaction project factor cluster group in the M initial risk transaction project factor cluster group sets, strategy extraction is performed on the M initial risk transaction project sets to determine the M risk transaction project sets; based on a one-to-one mapping relationship between risk transaction projects and risk transaction project basic data, the M initial risk transaction project basic data clusters are mapped and matched according to the M risk transaction project sets to obtain the M risk transaction project basic data clusters.

[0055] Further, based on the M initial risk transaction project factor sets, a factor multi-point aggregation analysis is performed to determine the M initial risk transaction project factor aggregation group sets, including:

[0056] Randomly extract M extracted initial risk transaction project factor sets from the M initial risk transaction project factor sets as M initial clustering factor sets; traverse the M initial clustering factor sets respectively to perform pairwise initial clustering factor similarity analysis to determine M similarity sets; determine whether the number of similarities in the M similarity sets whose similarities exceed a preset similarity is greater than or equal to a preset number, and if so, randomly extract M extracted initial risk transaction project factor sets from the M initial risk transaction project factor sets again to update the M initial clustering factor sets.

[0057] Furthermore, the method comprises:

[0058] If not, then taking the M initial clustering factor sets as clustering points respectively, a multi-point clustering analysis is performed on the M initial risk transaction project factor sets to determine the M initial risk transaction project factor clustering group sets.

[0059] Furthermore, the method comprises:

[0060] Randomly extracting M first initial clustering factors from the M initial clustering factor sets as M first clustering points without replacement; taking the M first clustering points as the centers of the M first clustering groups and taking a preset clustering step as the radius, constructing M first initial risk transaction project factor clustering groups based on the M initial clustering factor sets; performing edge outward diffusion of the M first initial risk transaction project factor clustering groups according to the preset clustering step to obtain M first diffused initial risk transaction project factor clustering groups; calculating the M first clustering densities of the M first initial risk transaction project factor clustering groups and the M first diffused clustering densities of the M first diffused initial risk transaction project factor clustering groups; when the M first When a diffusion clustering density is less than the M first clustering densities, the diffusion update is stopped, and the M first initial risk transaction project factor clustering groups are removed from the M initial risk transaction project factor sets to obtain updated M initial risk transaction project factor sets; M second initial clustering factors are randomly selected from the M initial clustering factor sets as M second clustering points without replacement, and clustering analysis is performed on the updated M initial risk transaction project factor sets to obtain M second initial risk transaction project factor clustering groups; based on the M initial clustering factor sets, multi-point clustering analysis is performed on the M initial risk transaction project factor sets to obtain the M initial risk transaction project factor clustering group sets.

[0061] Further, according to the factor amount in each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets, mapping is performed on the M initial risk transaction project sets to extract strategies, and the M risk transaction project sets are determined, including:

[0062] The number of factors of each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets is counted respectively to obtain M factor number sets; each factor number in the M factor number sets is divided by the sum of the factor numbers in the corresponding M factor number sets to obtain M extraction coefficient sets; a preset total extraction amount is obtained, and strategy extraction is performed on the M initial risk transaction project sets based on the M extraction coefficient sets and the preset total extraction amount to determine the M risk transaction project sets.

[0063] Furthermore, the method comprises:

[0064] Construct an adversarial risk identification network layer, wherein the adversarial risk identification network layer includes a generator and a discriminator; respectively use the M interference transaction project basic data clusters, the M risk transaction project sets and the M risk transaction project basic data clusters to train the generator and the discriminator until convergence, thereby obtaining the trained M adversarial risk identification network layers.

[0065] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0066] Furthermore, the first or second mentioned above may not only represent an order relationship, but may also represent a specific concept, and / or refer to the selection of multiple elements individually or in whole. Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application intends to include these modifications and variations.

Claims

1. A risk warning platform for rural property rights transactions, characterized by: The platform includes: A data collection module, used to collect basic data of the items to be traded, and obtain K basic data sets of K items to be traded, wherein the items to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1; A homogeneous aggregation module, used to perform homogeneous aggregation on the K basic data sets with the transaction mode as the clustering target, and determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction mode identifiers, and M is an integer less than or equal to K; A frequent item mining module, used to interact with the M transaction mode identifiers to perform risk transaction frequent item mining, and determine M risk transaction item sets and M risk transaction item basic data clusters; A network training module, used for acquiring M interference transaction project basic data clusters based on the M risk transaction project sets, training the adversarial risk identification network layer in combination with the M risk transaction project basic data clusters, and constructing M adversarial risk identification network layers; A risk identification module, configured to use the M counter-risk identification network layers to perform transaction risk identification on the M aggregated basic data clusters, and determine M counter-risk identification result sets; The warning information generation module is used to generate risk warning information when there are risky items to be traded in the M counter-risk identification result sets.

2. The risk early warning platform for rural property rights transactions according to claim 1, characterized in that: include: Based on the big data, relevant data of risky transaction projects are collected according to the M transaction mode identifiers to obtain M initial risky transaction project sets, M initial risky transaction project basic data clusters and M initial risky transaction project risk factor sets; Performing factor multi-point aggregation analysis based on the M initial risk transaction project factor sets to determine the M initial risk transaction project factor aggregation group sets; According to the factor amount in each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets, mapping is performed on the M initial risk transaction project sets for strategy extraction to determine the M risk transaction project sets; Based on a one-to-one mapping relationship between risk transaction projects and risk transaction project basic data, the M initial risk transaction project basic data clusters are mapped and matched according to the M risk transaction project sets to obtain the M risk transaction project basic data clusters.

3. The risk early warning platform for rural property rights transactions according to claim 2, characterized in that: Based on the M initial risk transaction project factor sets, a factor multi-point aggregation analysis is performed to determine the M initial risk transaction project factor aggregation group sets, including: Randomly extracting M extracted initial risk transaction project factor sets from the M initial risk transaction project factor sets as M initial aggregation factor sets; Traversing the M initial clustering factor sets respectively to perform pairwise initial clustering factor similarity analysis to determine M similarity sets; Determine whether the number of similarities in the M similarity sets whose similarities exceed the preset similarity is greater than or equal to the preset number. If so, randomly extract M extracted initial risk transaction project factor sets from the M initial risk transaction project factor sets to update the M initial clustering factor sets.

4. The risk early warning platform for rural property rights transactions according to claim 3, characterized in that: include: If not, then taking the M initial clustering factor sets as clustering points respectively, a multi-point clustering analysis is performed on the M initial risk transaction project factor sets to determine the M initial risk transaction project factor clustering group sets.

5. The risk early warning platform for rural property rights transactions according to claim 4, characterized in that: include: Randomly extracting M first initial clustering factors from the M initial clustering factor sets as M first clustering points without replacement; Taking the M first aggregation points as the centers of the M first aggregation groups and taking the preset aggregation step length as the radius, constructing M first initial risk transaction project factor aggregation groups based on the M initial aggregation factor sets; Performing edge outward diffusion on the M first initial risk transaction project factor clusters according to the preset clustering step length to obtain M first diffused initial risk transaction project factor clusters; Calculating the M first clustering densities of the M first initial risk transaction project factor clustering groups and the M first diffusion clustering densities of the M first diffusion initial risk transaction project factor clustering groups; When the M first diffusion aggregation densities are less than the M first aggregation densities, the diffusion update is stopped, and the M first initial risk transaction project factor aggregation groups are removed from the M initial risk transaction project factor sets to obtain an updated M initial risk transaction project factor sets; Again, randomly extracting M second initial clustering factors from the M initial clustering factor sets as M second clustering points without replacement, performing clustering analysis on the updated M initial risk transaction project factor sets, and obtaining M second initial risk transaction project factor clustering groups; Based on the M initial clustering factor sets, multi-point clustering analysis is performed on the M initial risk transaction project factor sets to obtain the M initial risk transaction project factor clustering group sets.

6. The risk warning platform for rural property rights transactions according to claim 2, characterized in that: According to the factor amount in each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets, mapping is performed on the M initial risk transaction project sets to extract strategies, and the M risk transaction project sets are determined, including: Respectively counting the number of factors of each initial risk transaction project factor cluster in the M initial risk transaction project factor cluster sets to obtain M factor number sets; Divide the number of each factor in the M factor number sets by the sum of the number of factors in the corresponding M factor number sets to obtain M extraction coefficient sets; A preset total amount of extraction is obtained, and strategy extraction is performed on the M initial risk transaction project sets based on the M extraction coefficient sets and the preset total amount of extraction to determine the M risk transaction project sets.

7. The risk early warning platform for rural property rights transactions according to claim 1, characterized in that: include: Constructing a confrontation risk identification network layer, wherein the confrontation risk identification network layer includes a generator and a discriminator; The generator and the discriminator are trained respectively using the M interference transaction project basic data clusters, the M risk transaction project sets and the M risk transaction project basic data clusters until convergence, so as to obtain the trained M adversarial risk identification network layers.

8. A risk warning method for rural property rights transactions, characterized in that: Based on the implementation of the risk warning platform for rural property rights transactions according to any one of claims 1 to 7, the method comprises: Perform basic data collection of the items to be traded to obtain K basic data sets of K items to be traded, wherein the items to be traded correspond to the basic data sets one by one, and K is an integer greater than or equal to 1; Taking the transaction mode as the clustering target, the K basic data sets are clustered in the same category to determine M aggregated basic data clusters, wherein the M aggregated basic data clusters have M transaction mode identifiers, where M is an integer less than or equal to K; Interacting the M transaction mode identifiers to perform risk transaction frequent item mining, and determining M risk transaction item sets and M risk transaction item basic data clusters; Based on the M risky transaction project sets, M interference transaction project basic data clusters are obtained, and the adversarial risk identification network layer is trained in combination with the M risky transaction project basic data clusters to construct the M adversarial risk identification network layers; Using the M counter-risk identification network layers to perform transaction risk identification on the M aggregated basic data clusters, and determining M counter-risk identification result sets; When there are risky items to be traded in the M counter-risk identification result sets, risk warning information is generated.

Images