Multi-platform-based non-performing asset information pushing method and multi-platform-based non-performing asset information pushing equipment

Abstract

The invention relates to the technical field of data information processing, solves the problem that a non-performing asset marketing scheme has significant defects in information symmetry and transaction efficiency in the prior art, and provides a multi-platform-based non-performing asset information pushing method and equipment. The method comprises the steps of obtaining recommendation information of each non-performing asset, pushing the recommendation information to each specified platform through a preset standardized interface, and calculating an asset popularity value of each non-performing asset according to behavior data of a user in each specified platform; pushing recommendation information of the interested unhealthy assets on the own channel platform according to the asset popularity value and / or the behavior data; and withdrawing the recommendation information of the corresponding non-performing assets in all the specified platforms through the preset standardized interface in response to the condition that the non-performing assets meet the preset withdrawing condition. According to the invention, the marketing efficiency of the unhealthy assets is improved, and the accuracy and correlation of the information are ensured.

Description

Technical Field

[0001] The present invention relates to the field of data information processing, and in particular to a method and device for pushing non-performing asset information based on multiple platforms. Background Art

[0002] Non-performing assets usually refer to asset losses caused by borrowers’ default or failure to perform contracts. Such assets mainly include non-performing loans, overdue accounts receivable, bankrupt corporate assets, etc. Due to the loss of value of non-performing assets, handling and disposing of non-performing assets has become an important task for financial institutions and investors. Effectively pushing and disposing of non-performing assets can not only reduce financial risks, but also help to realize asset value recovery and maximization.

[0003] At present, the external marketing of assets in the non-performing asset industry mainly relies on offline negotiations. Specifically, the transferor needs to reach a deal through face-to-face communication and negotiation with potential investors. Although this method can promote the conclusion of transactions to a certain extent, it also has obvious shortcomings. The existing offline negotiation method lacks an efficient and stable matching mechanism, resulting in information asymmetry between the transferor of non-performing assets and investors. The transferor cannot quickly find suitable investors, and investors find it difficult to obtain accurate information on specific assets. Investors often find it difficult to obtain accurate and reliable non-performing asset information, which affects the accuracy of their investment decisions. This lack of information has led to value mismatches in the market, and many potential high-quality non-performing assets cannot be reasonably valued and processed. Due to poor information and insufficient matching mechanisms, the transaction process of many non-performing assets is cumbersome and time-consuming, which limits the rapid marketing and circulation of non-performing assets. This inefficiency not only affects the market's activity, but also prevents the interests of investors and transferors from being realized in a timely manner.

[0004] In summary, the existing non-performing asset marketing schemes have significant defects in information symmetry and transaction efficiency. Summary of the invention

[0005] In view of this, the embodiments of the present invention provide a method and device for pushing non-performing asset information based on multiple platforms, so as to solve the problem that the non-performing asset marketing scheme in the prior art has significant defects in information symmetry and transaction efficiency.

[0006] In a first aspect, an embodiment of the present invention provides a method for pushing non-performing asset information based on multiple platforms, the method comprising:

[0007] Obtain the recommendation information of each non-performing asset, and push the recommendation information to each designated platform through a preset standardized interface, wherein the designated platform is divided into a self-owned channel platform and an external promotion platform;

[0008] Obtaining the first behavior data of the users in each of the designated platforms in response to the recommendation information, and calculating the asset heat value of each of the non-performing assets;

[0009] Pushing recommendation information of interested non-performing assets on the own channel platform according to the asset heat value and / or the second behavior data of the own channel platform;

[0010] In response to a non-performing asset satisfying a preset withdrawal condition, the recommendation information of the corresponding non-performing asset on all the designated platforms is withdrawn through the preset standardized interface.

[0011] Preferably, the obtaining of the first behavior data of the users in each of the designated platforms with respect to the recommendation information and calculating the asset heat value of each of the non-performing assets includes:

[0012] At every preset time period, the first behavior data of users on each of the designated platforms for the recommendation information is obtained through the preset standardized interface, wherein the self-owned channel platform includes an internal promotion platform and a service provider platform, the first behavior data on the external promotion platform and the internal promotion platform include the number of onlookers of the auction of each non-performing asset, and the first behavior data on the service provider platform includes the number of service provider solutions for each non-performing asset;

[0013] According to the number of auction spectators of all the non-performing assets, the total number of auction spectators on the external promotion platform and the internal promotion platform is obtained;

[0014] According to the number of service provider solutions for all the non-performing assets, the total number of solutions on the service provider platform is obtained;

[0015] Determine the behavioral data weight of each of the designated platforms according to the total number of auction spectators and the total number of plans;

[0016] A weighted calculation is performed based on the first behavior data of each designated platform and the corresponding behavior data weight to obtain the asset heat value of each non-performing asset.

[0017] Preferably, the pushing of recommendation information of the interested non-performing assets on the own channel platform according to the asset heat value and / or the second behavior data of the own channel platform includes:

[0018] Determine whether the user has logged into the own channel platform;

[0019] If logged in, determining the user's interested non-performing assets according to the login information, the second behavior data and the asset heat value;

[0020] If the user is not logged in, the user's interested non-performing assets are determined according to the asset heat value;

[0021] The recommendation information of the non-performing assets of interest is pushed to the proprietary channel platform through the preset standardized interface.

[0022] Preferably, the recommendation information includes the type of non-performing assets, and if logged in, determining the non-performing assets of interest to the user according to the login information, the behavior data and the asset heat value includes:

[0023] According to the login information of the user and the second behavior data, obtain the personal behavior data of each user on the self-owned channel platform, wherein the personal behavior data includes the number of first clicks of the user on each non-performing asset and the number of second clicks of the user on each type of non-performing assets;

[0024] Obtaining the user's interest level in all non-performing assets according to the first number of clicks, the second number of clicks, and the asset heat value;

[0025] The user's interested non-performing assets are determined according to the interest level value of each non-performing asset.

[0026] Preferably, obtaining the user's interest level value for all non-performing assets according to the first number of clicks, the second number of clicks, and the asset heat value includes:

[0027] Obtaining the first number of clicks, the second number of clicks, and initial weight values ​​of the asset heat value respectively;

[0028] Adjusting each of the initial weight values ​​according to the pre-built neural network model and the behavior data to obtain a target weight value;

[0029] Performing a weighted calculation according to the first number of clicks, the second number of clicks, the asset heat value, and the corresponding target weight value to obtain an initial interest level value for each non-performing asset;

[0030] The initial interest level values ​​are normalized to obtain the user's interest level values ​​for all non-performing assets.

[0031] Preferably, in response to a non-performing asset satisfying a preset withdrawal condition, withdrawing the recommendation information of the corresponding non-performing asset on all the designated platforms through the preset standardized interface includes:

[0032] Obtaining asset status information of each non-performing asset through the specified standard interface, wherein the asset status information includes: asset status change information and asset term information;

[0033] Determine whether the corresponding non-performing assets meet the preset withdrawal conditions according to the asset status change information and the asset term information;

[0034] When a non-performing asset meets the preset withdrawal conditions and / or receives a manual withdrawal instruction, a withdrawal instruction is sent to each of the designated platforms through the preset standard interface to withdraw the corresponding non-performing asset recommendation information.

[0035] Preferably, obtaining the recommendation information of each non-performing asset and pushing the recommendation information to each designated platform through a preset standardized interface includes:

[0036] Obtain initial multi-dimensional information of each non-performing asset based on the preset data source;

[0037] Preprocessing the initial multi-dimensional information of each non-performing asset to obtain target multi-dimensional information, wherein the preprocessing includes: data cleaning, data standardization and data verification;

[0038] Generate a multidimensional label for each non-performing asset according to the target multidimensional information and a preset rule engine, wherein the multidimensional label includes a disposal status label, an asset type label, a disposal difficulty label, and a historical auction label;

[0039] Generate recommendation information corresponding to each designated platform according to the multi-dimensional tag and the user portrait of each designated platform, recorded as platform recommendation information, wherein the user portrait is generated based on the first behavior data;

[0040] The platform recommendation information is pushed to the corresponding designated platform through the preset standardized interface.

[0041] Preferably, generating a multi-dimensional label for each non-performing asset according to the target multi-dimensional information and a preset rule engine includes:

[0042] Initializing the preset rule engine according to the preset non-performing asset business classification standard information to obtain a target rule engine;

[0043] Inputting the target multi-dimensional information into the target rule engine to obtain the initial multi-dimensional label of each non-performing asset;

[0044] Performing statistical analysis on the historical data of each non-performing asset according to the initial multi-dimensional label of each non-performing asset to obtain a representative index of each initial multi-dimensional label, wherein the representative index at least includes a mean and a variance;

[0045] According to the representative index of each initial multidimensional label and the corresponding representative index threshold, the corresponding initial multidimensional label is screened to obtain an intermediate multidimensional label;

[0046] Inputting the target multidimensional information and the corresponding intermediate multidimensional labels into a pre-constructed Bayesian network to obtain the joint probability between each of the intermediate multidimensional labels and the corresponding target multidimensional information;

[0047] According to the joint probability, determining whether there is a conflict between the intermediate multi-dimensional label and the corresponding target multi-dimensional information;

[0048] If there is a conflict, the target rule engine is adjusted according to the conflicting intermediate multidimensional labels, and the target multidimensional information is input into the target rule engine to obtain the initial multidimensional labels of each non-performing asset.

[0049] If there is no conflict, the initial multi-dimensional label of the non-performing asset is used as its final multi-dimensional label.

[0050] Preferably, in response to a non-performing asset meeting a preset withdrawal condition, after withdrawing the recommendation information of the corresponding non-performing asset on all the designated platforms through the preset standardized interface, the method further comprises:

[0051] According to the asset term information, the non-performing assets that are withdrawn beyond the preset selling term are obtained and recorded as overdue withdrawn assets;

[0052] According to the asset status information, obtaining a set of sold non-performing assets;

[0053] According to the multi-dimensional tags, obtaining the non-performing assets matching the overdue withdrawn assets from the sold non-performing assets set and recording them as target sold assets;

[0054] Obtaining recommendation information of each of the target sold assets on the designated platform for selling the target sold assets, and recording the information as sold recommendation information;

[0055] Acquire display parameters of the sold recommendation information, wherein the display parameters include display frequency, display time period and recommendation content of the recommendation information;

[0056] Performing data aggregation processing on the display parameters to generate aggregated display parameters;

[0057] Performing correlation analysis on the multi-dimensional labels of the overdue withdrawn assets and the multi-dimensional labels of the unsold non-performing assets to determine the similarity between each of the overdue withdrawn assets and the unsold non-performing assets;

[0058] Classifying the unsold non-performing assets according to the similarity to obtain a plurality of asset classification groups;

[0059] Matching the aggregated display parameters with each asset classification group to obtain optimized recommendation information for each asset classification group;

[0060] The optimized recommendation information is updated to each designated platform through the preset standardized interface to adjust the recommendation information of the currently unsold non-performing assets.

[0061] In a second aspect, an embodiment of the present invention provides an information push device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory, and when the computer program instructions are executed by the processor, the method of the first aspect in the above-mentioned embodiment is implemented.

[0062] In summary, the beneficial effects of the present invention are as follows:

[0063] The embodiment of the present invention provides a method and device for pushing information of non-performing assets based on multiple platforms. The method obtains the promotion information of each non-performing asset and pushes the promotion information to each designated platform through a preset standardized interface. The designated platform includes a self-owned channel platform and an external promotion platform. The standardized interface is used to ensure the rapid, accurate and consistent transmission of the non-performing asset promotion information, reduce human errors and delays in information transmission, and improve the efficiency of information release. In addition, covering multiple platforms broadens the exposure rate of assets and increases the interaction opportunities between the transferor and potential investors. By analyzing user behavior data, the market demand and investor interest of non-performing assets can be monitored and evaluated in real time to form a dynamic asset heat value. This mechanism helps the transferor to understand which assets are of concern in a timely manner, thereby optimizing marketing strategies and resource allocation. Pushing according to the asset heat value and user behavior data effectively improves the relevance and matching of information, making the promotion information more targeted. This not only improves the user's response rate to the promotion information, but also enhances the investor's sense of participation and satisfaction. By setting withdrawal conditions and performing real-time monitoring, the timeliness and accuracy of non-performing asset information are ensured. This flexible management approach can reduce the interference of outdated information to investors, thereby maintaining the credibility and effectiveness of platform information.

[0064] In summary, the method of the present invention not only improves the marketing efficiency of non-performing assets, but also ensures the accuracy and relevance of information, thereby better promoting the circulation and trading of non-performing assets and providing a more stable and efficient channel for the connection between transferors and investors. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the drawings required for use in the embodiment of the present invention will be briefly introduced below. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work, and these are all within the protection scope of the present invention.

[0066] Figure 1It is a flow chart of a method for pushing non-performing asset information based on multiple platforms according to an embodiment of the present invention.

[0067] Figure 2 It is another flow chart of a method for pushing non-performing asset information based on multiple platforms according to an embodiment of the present invention.

[0068] Figure 3 It is another flow chart of a method for pushing non-performing asset information based on multiple platforms according to an embodiment of the present invention.

[0069] Figure 4 This is another flow chart of a method for pushing non-performing asset information based on multiple platforms according to an embodiment of the present invention.

[0070] Figure 5 This is another flow chart of a method for pushing non-performing asset information based on multiple platforms according to an embodiment of the present invention.

[0071] Figure 6 It is a schematic diagram of the structure of an information push device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The features and exemplary embodiments of various aspects of the present invention will be described in detail below. In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are only configured to explain the present invention and are not configured to limit the present invention. For those skilled in the art, the present invention can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present invention by illustrating examples of the present invention.

[0073] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0074] It should be noted that all actions of acquiring signals, information or data in the present invention are performed in compliance with the corresponding data protection laws and policies of the location and with the authorization given by the owner of the corresponding device.

[0075] Example 1

[0076] See also Figure 1 The embodiment of the present invention provides a method for pushing non-performing asset information based on multiple platforms, the method comprising:

[0077] S1. Obtain promotion information of each non-performing asset, and push the promotion information to each designated platform through a preset standardized interface, wherein the designated platform is divided into a self-owned channel platform and an external promotion platform;

[0078] Specifically, promotional information refers to a detailed description of non-performing assets, including asset type, value, disposal method, market conditions and other information, so that potential investors can understand the basic situation of the assets. The preset standardized interface is a set of pre-set rules and protocols to ensure that information between different systems can be transmitted efficiently and seamlessly.

[0079] This step aims to increase the exposure of assets and attract the attention of potential investors by obtaining promotional information of non-performing assets and transmitting it to multiple platforms through standardized interfaces. Enhance the market liquidity of assets by disseminating information through multiple channels. Collect relevant information of non-performing assets through data collection tools, including basic information of assets, historical performance, market valuation, etc. Format the collected promotional information using preset standardized interfaces to ensure the consistency and readability of the information. Push the formatted promotional information to multiple designated platforms, which are divided into proprietary channel platforms and external promotion platforms. External promotion platforms mainly refer to third-party auction platforms, which are used to publish investment information of non-performing assets to attract potential buyers to bid and watch; proprietary channel platforms are internal or cooperative channels, which are used for precise push and internal asset management.

[0080] In a specific embodiment, the preset standardized interface is a RESTful interface, which is a Web service interface based on the REST (Representational State Transfer) architectural style and is widely used in modern Web applications and services. Its design principles aim to simplify and improve the scalability, maintainability and performance of Web services. In the non-performing asset recommendation system, the use of the RESTful interface can bring the following advantages: the recommendation information of non-performing assets can be efficiently transmitted between various designated platforms through the RESTful interface to ensure the real-time and accuracy of the data; resources such as the status, labels and recommendation information of non-performing assets can be flexibly managed and updated through the RESTful interface to support dynamically changing business needs. The RESTful interface can be seamlessly integrated with various client applications and external platforms, such as Web applications, mobile applications, etc., to facilitate the recommendation and trading of non-performing assets.

[0081] S2. Calculating the asset heat value of each of the non-performing assets according to the first behavior data of the users in each of the designated platforms in response to the recommendation information;

[0082] Specifically, the first behavioral data refers to the activity records on various platforms, including browsing history, click-through rate, number of favorites, comments, transaction behavior, etc. The asset heat value reflects the degree of user attention and interest in a certain non-performing asset in a specific period of time, and is an important indicator for evaluating asset market demand. By analyzing user behavioral data, the asset heat value is obtained in order to understand the market's interest in each non-performing asset in real time and help the transferor develop a more effective marketing strategy. Collect user behavior data on a designated platform, analyze the data regularly or in real time, and calculate the heat value of each non-performing asset. A weighted algorithm can be used to assign different weights to different behavioral data.

[0083] By calculating the asset heat value, the transferor can timely grasp the market demand dynamics and optimize the promotion strategy. For assets with higher heat, the promotion efforts can be increased. Conversely, the asset disposal method or pricing strategy can be adjusted to improve resource utilization efficiency.

[0084] S3. Pushing recommendation information of the non-performing assets of interest on the own channel platform according to the asset heat value and / or the second behavior data of the own channel platform;

[0085] Specifically, the first behavior data is the overall behavior data of users on all platforms, while the second behavior data is the behavior data from the own channel platform. Interested non-performing assets refer to non-performing assets that meet user preferences based on user behavior data and asset heat value. Based on the second behavior data of the own channel platform, the user's preferences are analyzed, and the asset heat value is combined to push the recommended information that meets the user's interests to the own channel platform in a timely manner to improve the relevance of information and user participation. Through precise push, the user experience is enhanced, and the user's attention and response rate to the recommended information are increased. This move can not only effectively facilitate transactions, but also improve users' trust and satisfaction with the platform and increase customer stickiness.

[0086] S4. In response to a non-performing asset satisfying a preset withdrawal condition, the recommendation information of the corresponding non-performing asset on all the designated platforms is withdrawn through the preset standardized interface.

[0087] Specifically, the preset withdrawal conditions refer to specific conditions that are set in advance to trigger the withdrawal of recommended information, such as assets that have been traded or asset status changes. It refers to the instructions issued to each designated platform to stop recommending a certain non-performing asset based on the withdrawal conditions. By timely withdrawing recommended information that is no longer applicable, users can avoid obtaining outdated or misleading information, and improve the platform's information transparency and reliability. This flexible management method will help maintain customer trust and improve asset circulation efficiency.

[0088] Preferably, the obtaining of the first behavior data of the users in each of the designated platforms with respect to the recommendation information and calculating the asset heat value of each of the non-performing assets includes:

[0089] S21. At every preset time period, obtaining the first behavior data of users on each of the designated platforms for the recommendation information through the preset standardized interface, wherein the self-owned channel platform includes an internal promotion platform and a service provider platform, the first behavior data on the external promotion platform and the internal promotion platform includes the number of onlookers of the auction of each of the non-performing assets, and the first behavior data on the service provider platform includes the number of service provider solutions for each of the non-performing assets;

[0090] Specifically, the preset time period is a fixed time interval set by the system, such as daily or weekly, which is used to regularly obtain the behavior data of the platform to ensure the timeliness and update of the data. The purpose of this step is to regularly collect the behavior data of each platform through the preset standardized interface. In this step, the external promotion platform and the internal promotion platform provide the number of onlookers during the auction, while the service provider platform provides the number of proposals submitted. The number of proposals submitted by the service provider portal refers to the number of proposals submitted for a specific non-performing asset through the service provider portal system, that is, one of the self-owned channel platforms. When the service provider views a non-performing asset in the portal system, it may submit a corresponding asset disposal plan or investment plan based on the characteristics and needs of the asset. For example: how to package the asset, investment strategy, buyer matching suggestions, debt disposal plan, etc. These plans represent the service provider's attention to and intention to deal with the non-performing asset. The number of submitted plans reflects the service provider's interest in the non-performing asset and the degree of active participation. If an asset can attract multiple service providers to submit disposal plans, it usually indicates that the asset has high attractiveness and value in the market. Different from ordinary viewing or clicking behaviors, submitting a proposal is a more in-depth and proactive behavioral data, which means that the service provider has a higher level of involvement in the asset and may consider further discussing cooperation or disposal operations.

[0091] S22. According to the number of auction spectators of all the non-performing assets, the total number of auction spectators on the external promotion platform and the internal promotion platform is obtained;

[0092] Specifically, the purpose of this step is to integrate the onlooker data of multiple promotion platforms to provide a comprehensive user interest index for heat calculation. The number of onlookers pulled from each platform is accumulated, and the total number of onlookers for external promotion platforms and internal promotion platforms is calculated separately. By integrating the onlooker data, the market attention of non-performing assets can be more comprehensively reflected, providing support for the accurate calculation of asset heat.

[0093] S23. According to the number of the service provider solutions for all the non-performing assets, obtain the total number of solutions of the service provider platform;

[0094] Specifically, this step is used to measure the service provider platform's interest in non-performing assets and the number of solutions they provide for asset disposal, adding an important dimension to the heat value calculation.

[0095] S24, determining the behavior data weight of each designated platform according to the total number of auction spectators and the total number of schemes;

[0096] S25. Perform a weighted calculation based on the first behavior data of each designated platform and the corresponding behavior data weight to obtain the asset heat value of each non-performing asset.

[0097] Specifically, in order to determine the weights of different platform data in asset popularity to reflect the influence of each platform in asset marketing, in one embodiment, the external promotion platform includes external platform 1 and external platform 2, and the proprietary channel platform includes the service provider platform and the internal platform 3, then weight 1: weight 2: weight 3: weight 4 = total number of onlookers on external platform 1: total number of onlookers on external platform 2: total number of service provider solutions: total number of onlookers on internal platform 3; after the above data are proportionally converted and normalized, the behavioral data weights of each of the designated platforms can be obtained;

[0098] Subsequently, the data and weights of different platforms will be weighted and calculated to obtain the final asset heat value, which will provide a reference for the subsequent asset promotion and package collection strategy. The asset heat of a non-performing asset = the number of onlookers on external platform 1 * weight 1 + the number of onlookers on external platform 2 * weight 2 + the number of service provider solutions * weight 3 + the number of onlookers on internal platform 3 * weight 4;

[0099] Through weighted calculation, the asset heat value can fully reflect the comprehensive attention paid by multiple platforms to non-performing assets. User behaviors on different platforms are reasonably integrated, making the calculation results more in line with the actual market situation. Through the calculation of heat value, the market's interest level in non-performing assets can be quantified, providing accurate data support for asset promotion and guiding enterprises to formulate more effective marketing strategies and package acquisition decisions.

[0100] Preferably, the pushing of recommendation information of the interested non-performing assets on the own channel platform according to the asset heat value and / or the behavior data includes:

[0101] S31, determining whether the user has logged in to the proprietary channel platform;

[0102] Specifically, the purpose of determining whether a user is logged in is to distinguish between personalized and general push strategies. The behavior data of logged-in users is more complete, and more accurate recommendation information can be provided; the data of logged-in users is limited, so push needs to be based on more extensive heat analysis. In other words, distinguishing between logged-in and logged-out states enables the system to adopt different push strategies based on the integrity of the user's data, thereby improving the accuracy of push information and user experience.

[0103] S32: If logged in, determining the user's interested non-performing assets according to the login information, the second behavior data and the asset heat value;

[0104] Specifically, login information includes but is not limited to user account data, such as name, role, historical browsing or investment records. The personalized behavior data of logged-in users is richer. By integrating the user's historical behavior and asset heat value, the non-performing assets that the user is interested in can be accurately identified. Based on the user's login information, the system extracts the user's second behavior data from the database, such as the type of assets recently viewed or collected. These behavior data are matched with the asset heat value, and the attractiveness of each non-performing asset to the user is calculated, and assets with high heat and in line with user preferences are recommended first. This improves the accuracy of the pushed content, making it easier for users to find assets of interest, and increases transaction conversion rates and user satisfaction.

[0105] Preferably, the recommendation information includes the type of non-performing assets, and if logged in, determining the non-performing assets of interest to the user according to the login information, the second behavior data and the asset heat value includes:

[0106] S321. Acquire personal behavior data of each user on the self-owned channel platform according to the login information of the user and the second behavior data, wherein the personal behavior data includes the number of first clicks of the user on each of the non-performing assets and the number of second clicks of the user on each type of the non-performing assets;

[0107] Specifically, the system extracts the user's historical behavior records from the second behavior data based on the user's login information on its own channel platform and records them as personal behavior data. The first click count and the second click count are extracted from the user's behavior data, and these data are classified into different assets and asset types. The first click count refers to the number of clicks made by the user on a specific non-performing asset, which is used to reflect the user's interest in a single asset. The second click count refers to the number of clicks made by the user on a specific type of non-performing asset (such as residential, commercial real estate, etc.), reflecting the user's overall preference for this type of asset.

[0108] S322: Obtain the user's interest level value for each of the non-performing assets according to the first number of clicks, the second number of clicks, and the asset heat value;

[0109] Specifically, the number of clicks on a certain asset by the user, the number of clicks on assets of the same type, and the popularity value of the asset are weighted to calculate the user's interest level in the asset, that is, the interest score for the asset.

[0110] Preferably, obtaining the user's interest level value for each of the non-performing assets according to the first number of clicks, the second number of clicks, and the asset heat value includes:

[0111] S3221. Obtaining initial weight values ​​of the first click quantity, the second click quantity, and the asset heat value respectively;

[0112] Specifically, the initial weight is used to characterize the influence of different data types, such as whether the number of clicks reflects user interest better than market popularity. The system assigns initial weight values ​​to the first click number, the second click number, and the asset popularity value through preset rules or historical data statistics.

[0113] S3222, adjusting each of the initial weight values ​​according to the pre-built neural network model and the behavior data to obtain a target weight value;

[0114] Specifically, the neural network model is an algorithm based on deep learning, which is used to analyze complex data and adaptively adjust model parameters, including multi-layer perceptron (MLP), convolutional neural network (CNN) or recurrent neural network (RNN), etc., which can be selected according to the data type and scenario. In this scenario, due to the need to deal with the weighted problem of user behavior data, a fully connected feedforward neural network is preferred. A fully connected feedforward neural network (FNN) is a neural network model widely used for structured data, especially in weighted calculations, feature mapping, and processing of complex relationships between data. User behavior data usually contains multi-dimensional information, such as the number of clicks, asset popularity, preference type, etc. These features may have nonlinear relationships with each other, and these relationships may not be captured by simple linear methods.

[0115] The system inputs the user's behavior data (number of clicks and asset popularity) into the neural network model for training. The model adjusts the initial weights based on the correlation between the data and generates more accurate target weights. This allows for dynamic optimization of weights, improves the calculation accuracy of the interest level value, and thus improves the recommendation effect.

[0116] S3223, performing a weighted operation according to the first number of clicks, the second number of clicks, the asset heat value, and the corresponding target weight value to obtain an initial interest level value for each of the non-performing assets;

[0117] Specifically, the influence of different data sources is combined through weighted calculation to obtain a preliminary interest level value. The following formula is used for calculation:

[0118] Interest level = first click number × weight 1 + second click number × weight 2 + asset popularity × weight 3

[0119] Through the above weighted calculation, different types of data are effectively integrated to ensure that the calculation results of the interest level value truly reflect the user's interests.

[0120] S3224. Normalize the initial interest level values ​​to obtain the user's interest level values ​​for each of the non-performing assets.

[0121] Specifically, the dimensional differences of the interest values ​​between different assets are normalized to make the results more comparable. After normalization, the interest values ​​between different assets can be directly compared, which helps to screen out the assets that users are most interested in.

[0122] S323: Determine the user's interested non-performing assets according to the interest level value.

[0123] Specifically, based on the interest level, the system can filter out assets that match the user's interests and generate personalized recommendations. The system sorts the normalized interest levels of all assets and selects several assets with the highest scores as the non-performing assets that the user is interested in, thereby providing accurate asset recommendations and improving user satisfaction and the transaction conversion rate of the platform.

[0124] S33, if the user is not logged in, determining the user's interested non-performing assets according to the asset heat value;

[0125] Specifically, there is insufficient behavioral data for logged-in users, so it is necessary to push based on overall market data (such as asset heat value) to ensure that the recommended assets are attractive to most users. The system generates push content based on the current platform's list of high-heat assets. If an asset has a high number of onlookers on both external and internal platforms, the system defaults to the asset as an asset of interest to users in the market and pushes it to logged-in users. Push content can use default filtering, such as "hot asset recommendations" or "recent high-attention assets" to attract users to browse further. Even if the user is not logged in, the platform can push high-value assets through market heat data to ensure a certain push effect and increase asset exposure and traffic.

[0126] S34. Push the recommendation information of the non-performing assets of interest to the proprietary channel platform through the preset standardized interface.

[0127] Finally, the system generates push content based on the matched asset information of interest and formats it into a push template supported by the platform. Through the preset interface, the push information is automatically sent to the front end of the own channel platform.

[0128] Preferably, in response to a non-performing asset satisfying a preset withdrawal condition, withdrawing the recommendation information of the corresponding non-performing asset on all the designated platforms through the preset standardized interface includes:

[0129] S41. Obtaining the asset status information of each of the non-performing assets through the designated standard interface, including the asset status change information and asset term information;

[0130] Specifically, asset status change information refers to the status change information of non-performing assets during their life cycle, such as when an asset changes from "for sale" to "sold"; asset term information refers to the effective promotion period of non-performing assets, such as the auction deadline, sales validity period, etc.;

[0131] The system requests data from each designated platform through a preset standardized interface to obtain real-time status information and maturity information of non-performing assets, ensuring that the system can obtain asset status changes in a timely manner, avoid expired or sold assets from continuing to be displayed on the platform, and improve the accuracy and timeliness of information management.

[0132] S42, judging whether the corresponding non-performing assets meet the preset withdrawal conditions according to the asset status change information and the asset term information;

[0133] Specifically, based on the status changes and expiration information of the assets, it is automatically determined which assets are no longer suitable for display, thereby triggering the withdrawal of the recommended information. The preset withdrawal conditions and predefined conditions for triggering withdrawal include but are not limited to the asset being sold, the expiration of the term, or no longer being recommended. If any condition is met, the asset is marked for withdrawal to avoid invalid or erroneous asset information from continuing to be displayed on the platform, thereby improving user experience and platform management efficiency.

[0134] S43. When a non-performing asset meets the preset withdrawal condition and / or receives a manual withdrawal instruction, a withdrawal instruction is issued to each of the designated platforms through the preset standard interface to withdraw the recommendation information of the corresponding non-performing asset.

[0135] Specifically, this step aims to execute the withdrawal task according to the preset conditions, while supporting the administrator to manually trigger the withdrawal when necessary to ensure the accuracy and flexibility of the information. When the system detects that an asset meets the preset withdrawal conditions or receives a manual withdrawal instruction, the system sends a withdrawal instruction to each platform through a standardized interface.

[0136] Preferably, obtaining the recommendation information of each non-performing asset and pushing the recommendation information to each designated platform through a preset standardized interface includes:

[0137] S11, obtaining initial multi-dimensional information of each of the non-performing assets according to a preset data source;

[0138] Specifically, the preset data source refers to multiple sources that the system has been connected to for obtaining non-performing asset information, such as financial institutions, asset management companies, third-party data service providers, etc. The initial multi-dimensional information is the original data about non-performing assets, including but not limited to basic information of assets such as asset name, amount, historical auction records, status information such as whether it has been disposed of, and other dimensions.

[0139] Specifically, comprehensive initial multi-dimensional information is collected from various data sources to ensure that the system has a sufficiently detailed understanding of the non-performing assets, laying the foundation for subsequent analysis and processing.

[0140] S12, preprocessing the initial multi-dimensional information to obtain target multi-dimensional information, wherein the preprocessing includes: data cleaning, data standardization and data verification;

[0141] Specifically, this step aims to improve data quality through preprocessing to ensure that the data is accurate during subsequent analysis, label generation, and recommendation information generation. First, data cleaning is used to remove invalid or duplicate data and correct errors in the data, such as null value completion, date format unification, currency unit conversion, etc., and then the data fields from different sources are standardized to ensure the consistency of the data structure. Finally, some incomplete information is completed using preset rules or external databases, and consistency checks are performed on multi-source information to ensure that the information is accurate. After the preprocessing is completed, the target multidimensional information obtained should have a standardized format and can be used for the next step of generating recommendation information.

[0142] S13, generating a multidimensional label for each of the non-performing assets according to the target multidimensional information and a preset rule engine, wherein the multidimensional label includes a disposal status label, an asset type label, a disposal difficulty label, and a historical auction label;

[0143] Specifically, the rule engine is a system that analyzes and classifies data based on predefined rules, such as generating "high value" or "low value" labels based on the amount of assets. The system analyzes the target multi-dimensional information according to the rule engine and generates multiple labels. Multi-dimensional labels provide a basis for asset classification and recommendation, improving the matching and accuracy of recommended information.

[0144] S14, generating recommendation information corresponding to each of the designated platforms according to the multi-dimensional tags and the user portraits of the designated platforms, recorded as platform recommendation information, wherein the user portraits are generated based on the first behavior data;

[0145] Specifically, based on the characteristics and behavior data of platform users, recommendation information that meets the user profile of their needs is generated to improve asset exposure and conversion rate. The system combines the multi-dimensional tags of assets and the user behavior data of the platform (such as browsing history, interest preferences) to generate recommendation information suitable for the platform. For example, if a platform user has a high interest in industrial real estate, the system will give priority to pushing relevant asset information, thereby generating personalized recommendation information for different platforms to improve the recommendation effect and user participation.

[0146] Preferably, generating recommendation information corresponding to each of the designated platforms according to the multi-dimensional tags and the behavior data of each of the designated platforms, recorded as platform recommendation information, includes:

[0147] S141. Extracting basic user characteristics and user preference characteristics of each designated platform according to the first behavior data of each designated platform, wherein the basic user characteristics include user age, user gender, and user device used, and the platform user preference characteristics include investment preference, asset category preference, purchasing power, and activity;

[0148] Specifically, basic user characteristics refer to static information such as age, gender, and device type that reflect user identity or usage habits. User preference characteristics are dynamic characteristics based on user behavior data that reflect their investment habits and purchasing tendencies. Through in-depth analysis of the behavior data of each designated platform, the extracted basic user characteristics and user preference characteristics will provide basic data support for subsequent asset recommendations. This feature extraction process can not only help the system better understand user needs, but also improve the personalization and accuracy of recommendation information, thereby effectively promoting the rapid marketing and circulation of non-performing assets.

[0149] S142, constructing a user profile of each of the designated platforms according to the basic characteristics of the user and the user preference characteristics;

[0150] Specifically, user portraits are comprehensive descriptions of users formed through data analysis and modeling techniques based on their basic characteristics and preference characteristics. They can help companies better understand user needs and thus achieve precision marketing in the promotion of non-performing assets.

[0151] For each user, a user profile containing various dimensions is generated based on their basic characteristics and preference characteristics. For example and not limitation, the user profile may contain the following fields:

[0152] Basic characteristics: user ID, age, gender, and type of device used.

[0153] Preference characteristics: investment preference (such as high risk or low risk), asset class preference (such as non-performing assets, real estate), purchasing power level (such as high, medium, low), activity level (such as high, normal, low).

[0154] By building user portraits based on basic user characteristics and user preference characteristics, we can achieve a deep understanding of user behavior, thereby improving the pertinence and effectiveness of non-performing asset recommendations. User portraits can not only help the platform achieve personalized recommendations in recommendation information, but also provide important decision-making basis for the optimization of overall market strategies.

[0155] S143, matching the user portrait with the multi-dimensional label of the non-performing assets, and obtaining the target non-performing assets of each designated platform;

[0156] Specifically, in this step, the goal is to identify the target NPAs on each designated platform by matching user portraits with the multi-dimensional labels of NPAs. This process involves accurate identification of user needs and effective docking with the characteristics of NPAs in order to push asset information that best meets user preferences;

[0157] The matching process can be accomplished by pre-establishing matching criteria, such as the user's investment preference must be consistent with the asset type of the non-performing asset; the user's purchasing power should be consistent with the valuation of the non-performing asset; the user's activity is associated with the disposal status of the asset, etc.; the algorithm can also be used to automatically match the user's profile with the non-performing asset label, first filtering out the corresponding type of non-performing assets based on the user's asset category preference. For example, if user A prefers real estate, only all real estate-related assets are considered. Each potential target asset is scored, and the scoring basis includes: the matching degree of investment preferences, the relative matching degree of purchasing power and asset valuation, the matching degree of disposal difficulty and the user's investment ability, and the comprehensive score of each asset is calculated and sorted according to the score, and the highest-scoring assets are selected as target non-performing assets.

[0158] S144, generating initial recommendation information according to the target multi-dimensional information and multi-dimensional labels of the target non-performing assets;

[0159] In this step, the goal is to use the multi-dimensional information and labels of the identified target non-performing assets to generate initial promotional information for subsequent display and promotion on the designated platform. First, it is necessary to clarify the relevant information and labels of the target non-performing assets, including asset type, disposal status, valuation information, and historical data. In addition, multi-dimensional labels also need to be extracted, such as disposal status labels, asset type labels, disposal difficulty labels, and historical auction labels.

[0160] The process of generating an initial pitch begins with extracting this key information. Build a clear, concise pitch by selecting the data that is most important to potential investors, such as asset type, valuation, and disposition status. For example, you could mention that this is a real estate asset for disposal, with a valuation of $500,000 and a history of three auctions. This information will be used to attract the investor's attention.

[0161] The generated promotional information also needs to be formatted to ensure that it can be displayed correctly on each platform. According to the requirements of different platforms, the appropriate promotional information template should be used for formatting. This may include setting the title, adding asset photos, layout of key information, and display of contact information and inquiry channels.

[0162] S145, determining the display form of the recommendation information of each designated platform according to the basic characteristics of the user;

[0163] Specifically, it is necessary to collect and analyze the basic characteristics of each designated platform user, including age, gender, and device used. These characteristics provide an important basis for the display format of the recommended information. For example, young users may prefer a simple and fashionable display style, while older users may prefer a traditional, easy-to-understand layout. The devices used by users, such as mobile phones, tablets, and computers, also affect the display format of the recommended information. On mobile devices, the display should be simple and clear, easy to slide and click; on desktop devices, more information and more complex layouts can be provided.

[0164] S146. Generate platform recommendation information corresponding to each of the designated platforms according to the initial recommendation information and the corresponding recommendation information display form.

[0165] Specifically, the initial recommendation information is organized according to the display format corresponding to the user characteristics, and the generated recommendation information is pushed to each designated platform through a preset standardized interface. The generated recommendation information can be A / B tested in a small range to evaluate the impact of different display formats on user engagement. Based on user feedback and behavior data, the content and display method of the recommendation information are continuously optimized to ensure maximum user attention and increase conversion rate.

[0166] S15. Pushing each platform recommendation information to the corresponding designated platform through the preset standardized interface.

[0167] Specifically, the system will finally send the generated referral information to each designated platform through a standardized interface. This process includes steps such as data packaging, sending instructions, and successful confirmation to ensure that the information reaches the platform accurately.

[0168] Preferably, generating a multi-dimensional label for each non-performing asset according to the target multi-dimensional information and a preset rule engine includes:

[0169] S131, initializing the preset rule engine according to the preset non-performing asset business classification standard information to obtain a target rule engine;

[0170] Specifically, the preset non-performing asset business classification standard information refers to pre-defined standard information used to describe the characteristics and classification of non-performing assets, such as asset types such as residential real estate, commercial real estate, and disposal methods such as auction and transfer. The system reads the preset classification standards and business rules and imports them into the rule engine. The initialization process includes loading the data model, rule configuration, and fine-tuning according to different industry practices and platform requirements. The initialized rule engine is more accurate and meets business needs.

[0171] S132, inputting the target multi-dimensional information into the target rule engine to obtain the initial multi-dimensional label of each of the non-performing assets;

[0172] Specifically, the system automatically passes the pre-processed target multi-dimensional information to the rule engine, which generates initial multi-dimensional labels according to the input information and preset logic. The labels include the basic characteristics and classification information of the assets. Through this step, the system can preliminarily screen and classify the assets to improve the accuracy and matching degree of asset recommendations.

[0173] S133, performing statistical analysis on the historical data of each of the non-performing assets according to the initial multi-dimensional labels, and obtaining representative indicators of each of the initial multi-dimensional labels, wherein the representative indicators at least include a mean and a variance;

[0174] Specifically, the system conducts statistical descriptive analysis on the historical data of non-performing assets (such as transaction records, market performance, user interaction data, etc.), including calculating representative indicators such as the mean and variance of each label. These indicators are used to evaluate the representativeness and stability of labels, thereby determining which labels can fully reflect the characteristics of non-performing assets.

[0175] S134, screening the corresponding initial multidimensional labels according to the representative index of each initial multidimensional label and the corresponding representative index threshold to obtain an intermediate multidimensional label;

[0176] Specifically, the system screens the initial multidimensional labels based on representative indicators and preset thresholds. Only those labels that meet the threshold requirements can be retained to obtain intermediate multidimensional labels. This screening process ensures that the generated labels are highly representative and accurate, and eliminates noisy and unreliable labels.

[0177] S135, inputting the target multi-dimensional information and the corresponding intermediate multi-dimensional labels into a pre-constructed Bayesian network to obtain a joint probability between each of the intermediate multi-dimensional labels and the corresponding target multi-dimensional information;

[0178] Specifically, the Bayesian network is a graph model based on probability theory, whose nodes represent variables and edges represent the dependencies between variables. In this scenario, the node variables include the target multidimensional information and the initial multidimensional label. In the model training phase, the system constructs a Bayesian network based on historical data, sets the dependencies between variables and the conditional probability table; the system converts the target multidimensional information and the initial multidimensional label into structured data that conforms to the Bayesian network input format, and inputs these data as observation variables into the corresponding nodes of the Bayesian network. The Bayesian network calculates the joint probability between the target multidimensional information and the initial multidimensional label through the conditional probability chain rule; the joint probability represents the probability of two or more events occurring at the same time. In this step, it is used to measure the mutual matching degree between the label and the multidimensional information, and to use the Bayesian network to determine whether there is a contradiction or deviation between the multidimensional label and the asset information. The system takes the target multidimensional information and the initial label as input and passes them to the Bayesian network for calculation. The Bayesian network analyzes the mutual dependence of each information and label, and outputs the joint probability between them to ensure the matching between the label and the asset information, and provide a basis for correcting potential erroneous labels.

[0179] S136, judging whether there is a conflict between the intermediate multi-dimensional label and the corresponding target multi-dimensional information according to the joint probability;

[0180] Specifically, conflict refers to the weak correlation or contradiction between the generated initial multidimensional labels and the target multidimensional information. The system will set a conflict threshold for the joint probability based on historical data, and use the conflict threshold to judge and detect logical inconsistencies between labels and information to avoid incorrect labels affecting subsequent analysis and recommendations;

[0181] S137: If there is a conflict, the target rule engine is adjusted according to the conflicting initial multi-dimensional labels, and the process returns to the step of inputting the target multi-dimensional information into the target rule engine to obtain the initial multi-dimensional labels of each of the non-performing assets;

[0182] Specifically, when there is a conflict, the system will adjust the conditional probability table or logic rules in the rule engine for the conflicting tags, and update the parameters of the rule engine through the feedback mechanism to make it more consistent with the actual situation and reduce the generation of erroneous tags. If a tag conflicts repeatedly, the system can trigger manual review and modify the model parameters. After adjustment, the system will return to step S132 and re-enter the target multi-dimensional information into the rule engine to generate a new initial multi-dimensional tag to ensure that the association between the tag and the asset information is more accurate, thereby improving the intelligence and adaptability of the system.

[0183] S138. If there is no conflict, the initial multi-dimensional label of the non-performing asset is used as its final multi-dimensional label.

[0184] Specifically, if the joint probability of all tags and asset information is higher than the preset threshold, it is determined that there is no conflict between the tag and the information. The system confirms that the currently generated intermediate multidimensional tags can truly reflect the characteristics of the assets, and directly stores the non-conflicting tags as the final multidimensional tags for subsequent analysis and recommendation. The final multidimensional tags ensure the accuracy of asset recommendation information, allowing investors to obtain real and reliable asset characteristics.

[0185] Through the above steps, the system can effectively generate platform promotion information corresponding to each designated platform, ensuring that the information is not only attractive but also meets the needs and expectations of users, thereby improving the display effect of non-performing assets and market circulation efficiency.

[0186] In one embodiment, in response to a non-performing asset satisfying a preset withdrawal condition, the recommendation information of the corresponding non-performing asset on all the designated platforms is withdrawn through the preset standardized interface, and the method further includes:

[0187] S05. According to the asset term information, obtain the non-performing assets that are withdrawn after the preset selling term, and record them as overdue withdrawn assets;

[0188] Specifically, the purpose of this step is to screen out non-performing assets that have been promoted on the platform for a period of time but have not been successfully sold. Identifying these assets helps to further analyze the shortcomings in their promotion process, thereby optimizing the promotion strategy. The system first queries the asset term information of each non-performing asset, and determines whether it exceeds the preset effective sales period by comparing the current time with the sales period. For those assets that have exceeded the period but have not been sold, they will be withdrawn from the promotion and recorded as "overdue withdrawn assets", effectively screening out assets that have not been sold within the scheduled time, facilitating subsequent cause analysis and promotion optimization, thereby improving the success rate of asset promotion.

[0189] S06. Acquire a set of sold non-performing assets according to the asset status information;

[0190] The purpose of this step is to identify those non-performing assets that have been successfully sold, collect their promotion data, and provide reference samples for subsequent promotion optimization. The system reads the asset status information of the non-performing assets and filters out the assets marked as sold. This process is usually implemented through database query to ensure that the latest sales status data is obtained.

[0191] S07. According to the multi-dimensional tags, obtain the non-performing assets matching the overdue withdrawn assets from the set of sold non-performing assets, and record them as target sold assets;

[0192] The system first searches for the most similar asset in the set of sold non-performing assets based on the multi-dimensional tags of each overdue withdrawn asset. This matching process can use similarity calculation algorithms, such as cosine similarity or Euclidean distance, to ensure that the sold assets found are as close as possible to the withdrawn assets in terms of attributes. By matching sold assets with similar tags, successful cases can be provided for the promotion strategy of overdue withdrawn assets, helping to improve the effectiveness of the promotion.

[0193] S08, obtaining the recommendation information of each of the target sold assets on the designated platform for selling the target sold assets, and recording the information as sold recommendation information;

[0194] The system extracts relevant recommendation information of the target sold assets from the recommendation records of the specified platform. This recommendation information is usually stored in the platform's database and may involve query operations on fields such as recommendation frequency, time period, and recommendation content. By analyzing the recommendation information of successfully sold assets, effective means and methods in the recommendation process can be summarized to provide data support for the optimization of the recommendation of similar assets.

[0195] S09, obtaining display parameters of the sold recommendation information, wherein the display parameters include display frequency, display time period and recommendation content of the recommendation information;

[0196] The system extracts various display parameters from the sold recommendation information, including display frequency (the number of times the recommendation information is displayed on the platform), display time period (the display arrangement of the recommendation information in a day or a specific time period) and recommendation content (the description method and graphic information of the recommendation). This step requires parsing and extracting the detailed fields in the recommendation information. Extracting these display parameters can provide rich information for subsequent data aggregation and analysis, thereby optimizing future recommendation strategies and increasing the recommendation effect of non-performing assets.

[0197] S010, performing data aggregation processing on the display parameters to generate aggregated display parameters;

[0198] Data aggregation processing refers to combining, classifying and statistically analyzing display parameters from multiple data sources to generate a representative result. Aggregate display parameters are comprehensive parameters obtained after processing multiple display parameters, reflecting the best display strategy.

[0199] The system performs statistical analysis on display parameters such as display frequency, display time period, and recommended content, and calculates indicators such as average display frequency and most frequently used time period to generate aggregated display parameters. This aggregation can use simple average calculation or weighted calculation to ensure that effective display modes are extracted. Through data aggregation, it is possible to summarize which display methods are most effective, so as to optimize them in future promotion strategies and improve the effectiveness of asset promotion.

[0200] S011, performing correlation analysis on the multi-dimensional labels of the overdue withdrawn assets and the multi-dimensional labels of the unsold non-performing assets, and determining the similarity between each of the overdue withdrawn assets and the unsold non-performing assets;

[0201] Specifically, by calculating the similarity, the feature similarity between overdue withdrawn assets and unsold assets can be identified to provide support for subsequent promotion optimization. Use machine learning or statistical analysis tools to compare the multi-dimensional labels of overdue withdrawn assets and unsold assets to calculate their similarity. Similarity calculation can be done using a variety of methods, such as Euclidean distance, Manhattan distance, or cosine similarity. Determining the similarity can help apply specific promotion strategies to more similar assets, thereby increasing the success rate of promotions.

[0202] S012. Classify the unsold non-performing assets according to the similarity to obtain multiple asset classification groups;

[0203] Specifically, asset classification groups divide unsold non-performing assets into different groups according to similarity, and the assets in each group have similar characteristics. The system divides the unsold assets into multiple classification groups based on the similarity calculated previously. The classification method can be based on clustering algorithms, such as K-means clustering or hierarchical clustering, to ensure that the characteristics of the assets in each classification group are as similar as possible. Classification can help the system adopt different promotion methods for different types of assets, thereby achieving refined operations and improving the success rate of promotion.

[0204] S013, matching the aggregate display parameters with each asset classification group to obtain optimized recommendation information for each asset classification group;

[0205] Specifically, the system matches the characteristics of each asset classification group with the aggregate display parameters to generate optimized recommendation information suitable for each group of assets. These recommendation information will include the frequency, time period and recommendation content suitable for display, so as to ensure that these assets can be promoted more effectively. Through personalized optimization of recommendation information, the exposure effect and conversion rate of assets can be significantly improved, which helps to sell assets as soon as possible.

[0206] S014. Update the optimized recommendation information to each designated platform through the preset standardized interface to adjust the recommendation information of the non-performing assets that have not yet been sold.

[0207] Specifically, the system updates the generated optimized recommendation information to each designated platform through a preset standardized interface. These updates usually include replacing old recommendation content, adjusting display time and frequency, etc. Data interaction is carried out through the interface to ensure that the recommendation information can be synchronized in a timely manner.

[0208] Example 2

[0209] In the scheme of Example 1, the asset heat value is obtained only by weighted calculation of the number of auction spectators in the behavioral data. This scheme can reflect the heat of non-performing assets to a certain extent, but it has the problems of single data source and relatively simplified weight setting. In order to improve the calculation accuracy of the asset heat value, this Example 2 is improved on the basis of Example 1.

[0210] In this embodiment, the obtaining of the first behavior data of the users in each of the designated platforms with respect to the recommendation information and the calculation of the asset heat value of each of the non-performing assets includes:

[0211] S1-21. At every preset time period, the first behavior data of the user in response to the recommendation information on each of the designated platforms is obtained through the preset standardized interface; wherein the owned channel platform includes an internal promotion platform and a service provider platform, the first behavior data on the external promotion platform and the internal promotion platform include positive behavior data and negative behavior data, and the first behavior data on the service provider platform includes the number of service provider solutions for each non-performing asset;

[0212] Specifically, the implementation process of this embodiment is similar to step S21 in Embodiment 1, except that, in this embodiment, the first behavior data under the external promotion platform and the internal promotion platform include positive behavior data and negative behavior data, wherein the positive behavior data includes:

[0213] Click behavior: records the number of times users click to view recommendation information, reflecting the user's direct interest in a specific asset; Browsing time: calculates the time users stay on the recommendation information page, indicating the user's deep attention; Sharing / forwarding behavior: calculates the number of times users share or forward the recommendation information, reflecting the social communication potential of the asset; Collection behavior: counts the number of users' collections of a certain asset, indicating the user's willingness to invest in the asset; Conversion behavior: such as signing up for an auction or submitting a purchase intention, used to measure the user's actual purchasing interest;

[0214] Negative behavior data includes but is not limited to: Bounce behavior, that is, users exit or jump to other pages within a short period of time, which usually means that the recommended information fails to attract the user's interest. Blocking operation: users block specific recommended information or non-performing asset categories, indicating that users do not want to see similar content again, and cancel attention, etc.;

[0215] Positive behavior reflects the user's active interest in bad assets, while negative behavior can show the user's rejection and disinterest. Combining these two types of data can better identify which assets are attractive to users and which are unpopular, thereby avoiding the recommendation of assets that users are clearly not interested in.

[0216] S1-22, removing abnormal behavior data from the first behavior data to obtain target behavior data;

[0217] Specifically, this step first performs a preliminary screening of the behavior data collected on different platforms to remove noise data. For example, data with a browsing time that is too short (e.g., less than 3 seconds) may be a missed click and can be excluded from the heat calculation. Abnormal behaviors (such as repeatedly clicking on a large amount of asset information in a short period of time) can be eliminated by setting rules. The elimination of abnormal behaviors can be based on the Z-Score method, filtering data that deviates from the mean by more than a set threshold, reducing the noise that affects the results, and thus obtaining the target behavior data;

[0218] S1-23, performing cluster analysis on the target behavior data of the external promotion platform and the internal promotion platform to obtain cluster analysis results;

[0219] In this step, the target behavior data of the external promotion platform and the internal promotion platform are clustered by using the K-means algorithm. The main goal is to divide users into different groups according to their behavior characteristics. First, multiple dimensions that can reflect the user's behavior characteristics are selected from the target behavior data as clustering features, including but not limited to the total number of clicks on specific non-performing assets by users, reflecting the user's attention to the assets. The time a user stays on the recommendation information page usually indicates the depth of the user's interest in the content; the number of shares or collections: the user's sharing or collection of the recommendation information indicates the user's affirmation of the asset; the number of negative behaviors: such as the number of operations such as quick exit, cancellation or blocking, indicating the user's negative tendency. The above features are standardized and the data of different dimensions are adjusted to the same scale to prevent some data from having too much impact on the clustering results. In the K-means clustering process, the Euclidean distance is used to calculate the similarity between users in each behavior feature. By comparing the differences in features such as the number of clicks and the length of stay, the similarity of user behavior is measured. Through multiple experiments and using methods such as the elbow method or the silhouette coefficient to select a suitable K value, ensure that users are reasonably divided into multiple groups. Generally, K can be selected between 3 and 5, such as "high interaction group", "low interaction group", "high potential investment group", etc. After determining the K value, the K-means algorithm is used to cluster the user's behavior data. During the operation of the algorithm, each user will be classified into the group with the most similar behavior characteristics.

[0220] S1-24. According to the cluster analysis result, users of the external promotion platform and the internal promotion platform are divided to obtain a number of user groups and corresponding user behavior feature labels;

[0221] Specifically, users are assigned to different groups based on the previous K-means clustering analysis results. Each group represents a group of users with high similarity in behavioral data. A corresponding behavioral feature label is assigned to each user group to identify the main behavioral characteristics and preferences of the group. Feature labels include not only positive behavioral data (such as click-through rate, dwell time, number of shares, etc.), but also negative behavioral data (such as quick exits, cancel operations, etc.). The behavioral preferences of each group can be further analyzed based on feature labels. For example, some user groups may prefer to click on distressed assets with lower prices, or show significant interest in a specific type of asset (such as real estate). This information can be recorded as a behavioral preference label.

[0222] S1-25, performing a weighted operation based on the user behavior feature label and the target behavior data to obtain the asset heat value of each non-performing asset;

[0223] Specifically, the corresponding weighting coefficient is set according to the different behavioral feature labels of each user. These weights can be based on the judgment of the actual interest in non-performing assets and the investment potential. Generally, the weight of positive behaviors (such as click-through rate, dwell time) is greater than that of negative behaviors (such as bounce rate), and the weight of high-interaction users is greater than that of low-interaction users. For example, the weight of the high-interaction group can be set between 1.2-1.5 to highlight the contribution of their behavior to the popularity of the asset. The weight of the low-interaction group can be set between 0.5-0.8 to reduce the impact of the group on the popularity. For negative behavior labels (such as quick bounces), negative weights can be set to reduce the contribution of users who are not interested in non-performing assets. Multiply the standardized behavioral data by the weighting coefficient to reflect the contribution of different behaviors and groups to the popularity of the asset.

[0224] Specifically, the behavior patterns of different groups may be different, so the above weights can be adjusted according to the results of cluster analysis. For example, if a group shows high activity, frequent clicks and long stays, the weight of their positive behavior can be appropriately increased to highlight their interest level. Specifically, the following rules can be followed: the weight of positive behavior is increased for highly active groups, and the weight of negative behavior is adjusted for low-activity groups to capture rejection tendencies.

[0225] After adding negative behaviors, the solution can more accurately identify users' interests and rejection tendencies. Based on the fusion weight calculation of positive and negative behaviors, combined with judgment and cluster analysis, it can effectively screen out uninterested users, reduce invalid push notifications, and ensure that recommended content is more in line with users' actual needs.

[0226] By way of example and not limitation, for users with more negative tendencies, the push frequency can be appropriately reduced to avoid users feeling bored; for users with a high frequency of blocking or pop-up, try to push different types of non-performing assets to avoid unchanging recommended content; for users with obvious positive tendencies and less negative behavior, further refine the push content to make the recommended information more in line with their needs and preferences.

[0227] Example 3

[0228] In addition, combined Figure 1 The method for pushing non-performing asset information based on multiple platforms described in the embodiment of the present invention can be implemented by an information pushing device. Figure 6 A schematic diagram of the hardware structure of an information push device provided by an embodiment of the present invention is shown.

[0229] The information push device may include a processor and a memory storing computer program instructions.

[0230] Specifically, the processor may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present invention.

[0231] The memory may include a large capacity memory for data or instructions. By way of example and not limitation, the memory may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory may include a removable or non-removable (or fixed) medium. Where appropriate, the memory may be inside or outside a data processing device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM) or a flash memory or a combination of two or more of these.

[0232] The processor implements any one of the multi-platform based non-performing asset information push methods in the above embodiments by reading and executing computer program instructions stored in the memory.

[0233] In one example, the information push device may further include a communication interface and a bus. Figure 6As shown, the processor 401, the memory 402, and the communication interface 403 are connected via a bus 410 and communicate with each other.

[0234] The communication interface is mainly used to implement communication between the modules, devices, units and / or equipment in the embodiments of the present invention.

[0235] The bus includes hardware, software or both, and the parts of the information push device are coupled to each other. For example, but not limitation, the bus may include accelerated graphics port (AGP) or other graphics buses, enhanced industrial standard architecture (EISA) bus, front-side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In suitable cases, the bus may include one or more buses. Although the embodiment of the present invention describes and shows a specific bus, the present invention considers any suitable bus or interconnection.

[0236] Example 4

[0237] In addition, in combination with the non-performing asset information push method based on multiple platforms in the above embodiments, the embodiments of the present invention can provide a computer-readable storage medium for implementation. The computer-readable storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any one of the non-performing asset information push methods based on multiple platforms in the above embodiments is implemented.

[0238] In summary, the method and device for pushing information on non-performing assets based on multiple platforms and the storage medium provided by the embodiment of the present invention obtain the promotion information of each non-performing asset, and push the promotion information to each designated platform through a preset standardized interface, wherein the designated platform includes a self-owned channel platform and an external promotion platform. Through the use of the standardized interface, the fast, accurate and consistent transmission of the promotion information of the non-performing assets is ensured, the human errors and delays in the information transmission are reduced, and the efficiency of information release is improved. In addition, covering multiple platforms broadens the exposure rate of assets and increases the interaction opportunities between the transferor and potential investors. By analyzing the user behavior data, the market demand and investor interest of non-performing assets can be monitored and evaluated in real time to form a dynamic asset heat value. This mechanism helps the transferor to understand which assets are of concern in a timely manner, thereby optimizing marketing strategies and resource allocation. Pushing according to the asset heat value and user behavior data effectively improves the relevance and matching of information, making the promotion information more targeted. This not only improves the user's response rate to the promotion information, but also enhances the investor's sense of participation and satisfaction. By setting withdrawal conditions and conducting real-time monitoring, the timeliness and accuracy of non-performing asset information are ensured. This flexible management approach can reduce the interference of outdated information to investors, thereby maintaining the credibility and effectiveness of platform information.

[0239] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0240] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present invention are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0241] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0242] The above is only a specific implementation of the present invention. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be covered within the protection scope of the present invention.

Claims

1. A method for pushing non-performing asset information based on multiple platforms, characterized in that: The method comprises: Obtaining promotion information of each non-performing asset, and pushing the promotion information to each designated platform through a preset standardized interface, wherein the designated platform is divided into a self-owned channel platform and an external promotion platform; Obtaining first behavior data of users in each of the designated platforms in response to the recommendation information, and calculating the asset heat value of each of the non-performing assets; Pushing recommendation information of interested non-performing assets on the own channel platform according to the asset heat value and / or the second behavior data of the own channel platform; In response to a non-performing asset satisfying a preset withdrawal condition, the recommendation information of the corresponding non-performing asset on all the designated platforms is withdrawn through the preset standardized interface.

2. The method for pushing non-performing asset information based on multiple platforms according to claim 1, characterized in that: The obtaining of the first behavior data of the users in each of the designated platforms with respect to the recommendation information and calculating the asset heat value of each of the non-performing assets includes: At every preset time period, the first behavior data of users on each of the designated platforms for the recommendation information is obtained through the preset standardized interface, wherein the self-owned channel platform includes an internal promotion platform and a service provider platform, the first behavior data on the external promotion platform and the internal promotion platform include the number of onlookers of the auction of each non-performing asset, and the first behavior data on the service provider platform includes the number of service provider solutions for each non-performing asset; According to the number of auction spectators of all the non-performing assets, the total number of auction spectators on the external promotion platform and the internal promotion platform is obtained; According to the number of service provider solutions for all the non-performing assets, the total number of solutions on the service provider platform is obtained; Determine the behavioral data weight of each of the designated platforms according to the total number of auction spectators and the total number of plans; A weighted calculation is performed based on the first behavior data of each designated platform and the corresponding behavior data weight to obtain the asset heat value of each non-performing asset.

3. The method for pushing non-performing asset information based on multiple platforms according to claim 2, characterized in that: The method of pushing recommendation information of the interested non-performing assets on the own channel platform according to the asset heat value and / or the second behavior data of the own channel platform includes: Determine whether the user has logged into the own channel platform; If logged in, determining the user's interested non-performing assets according to the login information, the second behavior data and the asset heat value; If the user is not logged in, the user's interested non-performing assets are determined according to the asset heat value; The recommendation information of the non-performing assets of interest is pushed to the proprietary channel platform through the preset standardized interface.

4. The method for pushing non-performing asset information based on multiple platforms according to claim 3 is characterized in that: The recommendation information includes the type of non-performing assets. If logged in, determining the non-performing assets of interest to the user according to the login information, the behavior data and the asset heat value includes: According to the login information of the user and the second behavior data, obtain the personal behavior data of each user on the self-owned channel platform, wherein the personal behavior data includes the number of first clicks of the user on each non-performing asset and the number of second clicks of the user on each type of non-performing assets; Obtaining the user's interest level in all non-performing assets according to the first number of clicks, the second number of clicks, and the asset heat value; The user's interested non-performing assets are determined according to the interest level value of each non-performing asset.

5. The method for pushing non-performing asset information based on multiple platforms according to claim 4 is characterized in that: The obtaining, according to the first number of clicks, the second number of clicks, and the asset heat value, the user's interest level value for all non-performing assets includes: Obtaining the first number of clicks, the second number of clicks, and initial weight values ​​of the asset heat value respectively; Adjusting each of the initial weight values ​​according to the pre-built neural network model and the behavior data to obtain a target weight value; Performing a weighted calculation according to the first number of clicks, the second number of clicks, the asset heat value, and the corresponding target weight value to obtain an initial interest level value for each non-performing asset; The initial interest level values ​​are normalized to obtain the user's interest level values ​​for all non-performing assets.

6. The method for pushing non-performing asset information based on multiple platforms according to any one of claims 1 to 5, characterized in that: In response to a non-performing asset satisfying a preset withdrawal condition, withdrawing the recommendation information of the corresponding non-performing asset on all the designated platforms through the preset standardized interface includes: Obtaining asset status information of each non-performing asset through the specified standard interface, wherein the asset status information includes: asset status change information and asset term information; Determine whether the corresponding non-performing assets meet the preset withdrawal conditions according to the asset status change information and the asset term information; When a non-performing asset meets the preset withdrawal conditions and / or receives a manual withdrawal instruction, a withdrawal instruction is sent to each of the designated platforms through the preset standard interface to withdraw the corresponding non-performing asset recommendation information.

7. The method for pushing non-performing asset information based on multiple platforms according to claim 6, characterized in that: Obtain the recommendation information of each non-performing asset, and push the recommendation information to each designated platform through a preset standardized interface, including: Obtain initial multi-dimensional information of each non-performing asset based on the preset data source; Preprocessing the initial multi-dimensional information of each non-performing asset to obtain target multi-dimensional information, wherein the preprocessing includes: data cleaning, data standardization and data verification; Generate a multidimensional label for each non-performing asset according to the target multidimensional information and a preset rule engine, wherein the multidimensional label includes a disposal status label, an asset type label, a disposal difficulty label, and a historical auction label; Generate recommendation information corresponding to each designated platform according to the multi-dimensional tag and the user portrait of each designated platform, recorded as platform recommendation information, wherein the user portrait is generated based on the first behavior data; The platform recommendation information is pushed to the corresponding designated platform through the preset standardized interface.

8. The method for pushing non-performing asset information based on multiple platforms according to claim 7, characterized in that: The step of generating a multi-dimensional label for each non-performing asset according to the target multi-dimensional information and a preset rule engine includes: Initializing the preset rule engine according to the preset non-performing asset business classification standard information to obtain a target rule engine; Inputting the target multi-dimensional information into the target rule engine to obtain the initial multi-dimensional label of each non-performing asset; Performing statistical analysis on the historical data of each non-performing asset according to the initial multi-dimensional label of each non-performing asset to obtain a representative index of each initial multi-dimensional label, wherein the representative index includes at least a mean and a variance; According to the representative index of each initial multidimensional label and the corresponding representative index threshold, the corresponding initial multidimensional label is screened to obtain an intermediate multidimensional label; Inputting the target multidimensional information and the corresponding intermediate multidimensional labels into a pre-constructed Bayesian network to obtain the joint probability between each intermediate multidimensional label and the corresponding target multidimensional information; According to the joint probability, determining whether there is a conflict between the intermediate multi-dimensional label and the corresponding target multi-dimensional information; If there is a conflict, the target rule engine is adjusted according to the conflicting intermediate multidimensional labels, and the target multidimensional information is input into the target rule engine to obtain the initial multidimensional labels of each non-performing asset. If there is no conflict, the initial multi-dimensional label of the non-performing asset is used as its final multi-dimensional label.

9. The method for pushing non-performing asset information based on multiple platforms according to claim 8, characterized in that: The method further comprises, in response to a non-performing asset satisfying a preset withdrawal condition, withdrawing the recommendation information of the corresponding non-performing asset on all the designated platforms through the preset standardized interface: According to the asset term information, the non-performing assets that are withdrawn beyond the preset selling term are obtained and recorded as overdue withdrawn assets; According to the asset status information, obtaining a set of sold non-performing assets; According to the multi-dimensional tags, obtaining the non-performing assets matching the overdue withdrawn assets from the sold non-performing assets set and recording them as target sold assets; Obtaining recommendation information of each of the target sold assets on the designated platform for selling the target sold assets, and recording the information as sold recommendation information; Acquire display parameters of the sold recommendation information, wherein the display parameters include display frequency, display time period and recommendation content of the recommendation information; Performing data aggregation processing on the display parameters to generate aggregated display parameters; Performing correlation analysis on the multi-dimensional labels of the overdue withdrawn assets and the multi-dimensional labels of the unsold non-performing assets to determine the similarity between each of the overdue withdrawn assets and the unsold non-performing assets; Classifying the unsold non-performing assets according to the similarity to obtain a plurality of asset classification groups; Matching the aggregated display parameters with each asset classification group to obtain optimized recommendation information for each asset classification group; The optimized recommendation information is updated to each designated platform through the preset standardized interface to adjust the recommendation information of the currently unsold non-performing assets.

10. An information push device, characterized in that: include: At least one processor, at least one memory and computer program instructions stored in the memory, when the computer program instructions are executed by the processor, implement the multi-platform based non-performing asset information push method as described in any one of claims 1-9.

Images