A bidirectional feedback type special asset intelligent recommendation method and system fusing risk assessment factors

Abstract

The application discloses a bidirectional feedback type special asset intelligent recommendation method and system fusing risk assessment factors. The method realizes accurate, safe and efficient recommendation of special assets by constructing an asset and investor portrait containing risk factors, fusing risk assessment factors, intelligent matching and recommendation algorithms, a bidirectional feedback mechanism and model optimization, and recommendation result explanation and risk prompt. The application can significantly improve recommendation accuracy and safety, shorten the decision-making cycle, optimize asset liquidity, enhance user trust, and realize system self-adaptation and intelligence.

Description

Technical Field

[0001] This invention belongs to the field of information technology, specifically involving artificial intelligence, machine learning, recommendation systems, and risk management technology, and is particularly applicable to the intelligent matching and personalized recommendation of assets and investors in the field of special assets (non-performing assets). Background Technology

[0002] In the specialized asset trading market, investors face a massive amount of asset information. However, due to the complexity, non-standardization, and potentially high risks of these assets, traditional recommendation systems (such as those based on collaborative filtering or content recommendation) struggle to effectively meet the unique needs of specialized asset investment. Existing recommendation systems typically focus on matching user interests while neglecting the crucial aspects of risk assessment and management capabilities in specialized asset investment. Furthermore, traditional recommendation systems often lack effective two-way feedback mechanisms, making it difficult to dynamically adjust and optimize based on actual transaction results and investors' risk tolerance. This results in insufficient accuracy and practicality of the recommendations, requiring investors to invest significant effort in risk identification and due diligence, leading to low transaction efficiency. Summary of the Invention

[0003] This invention aims to address the problem that existing special asset recommendation systems lack risk assessment integration and two-way feedback mechanisms, and provides a two-way feedback intelligent recommendation method and system for special assets that integrates risk assessment factors, so as to achieve accurate, safe and efficient recommendation of special assets. Technical solution

[0004] This invention provides a two-way feedback intelligent recommendation method for special assets that integrates risk assessment factors, comprising the following steps: 1. Asset and Investor Profile Construction: Asset Profile Construction: Based on standardized special asset data, a multi-dimensional asset profile is constructed, including asset type, geographical location, value range, ownership status, mortgage status, litigation status, disposal difficulty, and potential returns. Simultaneously, external or self-developed risk assessment models are integrated to generate risk scores and risk tags for each asset. Investor Profile Construction: Data on investor browsing behavior, search history, collection preferences, transaction records, financial strength, risk appetite, regional restrictions, and disposal experience on the platform are recorded and analyzed to construct a multi-dimensional investor profile. Specifically, the investor profile includes their acceptance of assets with different risk levels.

[0005] 2. Risk Assessment Factor Integration: Risk scores and labels from asset profiles are used as core factors and matched and integrated with risk preferences and risk tolerance from investor profiles. This ensures that recommended assets are aligned with the investor's risk tolerance.

[0006] 3. Intelligent Matching and Recommendation Algorithm: Design and implement an intelligent matching algorithm that integrates content matching, collaborative filtering, deep learning, and other technologies. This algorithm comprehensively considers various features of asset profiles (including risk factors) and investor profiles (including risk preferences) to calculate the matching degree. When generating a recommendation list, the recommendation algorithm prioritizes assets whose risk levels match the investor's risk preferences, and, under the same matching degree, prioritizes recommending assets with lower risk or easier disposal.

[0007] 4. Two-way Feedback Mechanism and Model Optimization: Explicit Investor Feedback: Collecting investor behavior data such as clicks, favorites, purchase requests, and transactions related to recommended assets, as well as explicit feedback such as satisfaction ratings and reasons for disinterest in the recommendation results. Implicit Transaction Feedback: Collecting transaction result data such as the actual transaction cycle, transaction price, disposal success rate, and subsequent disposal status of assets as implicit feedback. Iterative Model Optimization: Inputting explicit and implicit feedback data into the recommendation algorithm model, and dynamically adjusting the parameters of the asset risk assessment model, investor risk preference model, and intelligent matching algorithm through mechanisms such as reinforcement learning or online learning, to achieve continuous optimization and self-adaptation of the recommendation system.

[0008] 5. Explanation of Recommendations and Risk Warnings: The recommendations include the asset's risk level, key risks, and explanations of the reasons for the recommendation, helping investors understand the logic behind the recommendation and make informed decisions. For high-risk assets, clear risk warnings and necessary due diligence recommendations are provided.

[0009] This invention also provides a two-way feedback intelligent recommendation system for special assets that integrates risk assessment factors, comprising: a profile construction module for constructing profiles of special assets (including risk scores) and investor profiles (including risk preferences); a risk fusion module for matching and fusing asset risk factors with investor risk preferences; an intelligent recommendation module for executing an intelligent matching algorithm to generate a personalized recommendation list; a feedback collection module for collecting explicit feedback from investors and implicit feedback from transaction results; a model optimization module for iteratively optimizing the recommendation algorithm model based on feedback data; and an explanation and prompt module for providing explanations of the recommendation results and risk prompts. Beneficial effects

[0010] 1. Improve Recommendation Accuracy and Security: Deeply integrate risk assessment factors to ensure that recommendations not only align with investors' interests but also their risk tolerance, significantly improving accuracy and security. 2. Shorten Decision-Making Cycle: Investors can quickly discover assets that match their risk appetite and investment strategies, reducing screening and due diligence time and accelerating the decision-making process. 3. Optimize Asset Liquidity: Through precise recommendations, assets are pushed to the most suitable investors, improving the success rate and turnover efficiency of special assets. 4. Enhance User Trust: Transparent risk warnings and explanations of recommendation reasons, along with continuously optimized recommendation performance, help build investor trust in the platform. 5. Achieve System Adaptability and Intelligence: A two-way feedback mechanism enables the recommendation system to learn and optimize itself based on market changes and user behavior, continuously improving recommendation performance. Attached Figure Description

[0011] Figure 1 is a schematic diagram of the overall system architecture according to an embodiment of the present invention. Figure 2 is a flowchart of the recommended process according to an embodiment of the present invention. Figure 3 is a schematic diagram of the algorithm model logic according to an embodiment of the present invention. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0013] Please refer to Figure 1, which illustrates the overall architecture of a bidirectional feedback intelligent recommendation system for special assets that integrates risk assessment factors, according to an embodiment of the present invention. This system architecture includes a user layer, an application layer, a core service layer, a data layer, and an infrastructure layer. The user layer interacts with the application layer through the recommendation system front-end. The application layer provides recommendation service interfaces and is responsible for displaying recommendation results and collecting user behavior data. The core service layer is the core of the invention, including an investor profile building module, an asset profile building module, a risk assessment module, a risk fusion module, an intelligent matching and recommendation module, a feedback collection module, and a recommendation result interpretation and risk warning module. The data layer stores investor profiles, asset profiles, risk models, recommendation algorithm models, and feedback data. The infrastructure layer provides a big data platform, a cloud computing platform, and an AI algorithm platform as underlying support.

[0014] Please refer to Figure 2, the recommendation flowchart of this embodiment of the invention, which illustrates the complete process from asset data entry to model iteration and optimization. First, after "standardized asset data entry," "asset profile construction" is performed. Simultaneously, after "investor registration / login," "investor behavior data collection" and "investor profile construction" are conducted. The asset profile and investor profile are then fused using "risk assessment factors" and enter the "intelligent matching and recommendation algorithm" to generate a "personalized recommendation list," which includes risk warnings when displaying the "recommendation results." The investor's "explicit feedback" and "implicit feedback on transaction results" are collected and used for "model iteration and optimization," thereby feeding back into the asset profile, investor profile, risk assessment factor fusion, and intelligent matching and recommendation algorithm, forming a closed-loop optimization.

[0015] Please refer to Figure 3, a schematic diagram of the algorithm model logic of this embodiment of the invention, which illustrates the core working mechanism of the recommendation algorithm. In the "input layer," "standardized asset features" are transformed into "asset risk scores," and "investor preference features" are transformed into "investor risk preferences." In the "matching layer," assets and investors are dually matched through a "risk matching module" and an "interest matching module," and the final matching result is obtained through "comprehensive matching degree calculation." In the "output layer," a "recommendation list" is generated, and "explanation of recommendation results" and "risk warnings" are provided. Simultaneously, the "feedback and optimization" layer optimizes the "risk matching module," "interest matching module," and "model parameters" by collecting "investor behavior feedback" and "transaction result feedback," thereby continuously improving the accuracy and effectiveness of the recommendations.

Claims

1. A two-way feedback intelligent recommendation method for special assets that integrates risk assessment factors, characterized in that, Includes the following steps: Asset and Investor Profile Construction: Based on standardized special asset data, a multi-dimensional asset profile is constructed, including asset type, geographical location, value range, ownership status, mortgage status, litigation status, disposal difficulty, and potential returns. Simultaneously, external or self-developed risk assessment models are integrated to generate risk scores and risk tags for each asset. Data on investor browsing behavior, search history, collection preferences, transaction records, financial strength, risk appetite, regional restrictions, and disposal experience on the platform are recorded and analyzed to construct a multi-dimensional investor profile, which includes the investor's acceptance of assets with different risk levels. Risk assessment factor fusion: Risk scores and risk tags from asset profiles are used as core factors and matched and integrated with risk preferences and risk tolerance from investor profiles to ensure that recommended assets are in line with the investor's risk tolerance. Intelligent matching and recommendation algorithm: An intelligent matching algorithm integrating content matching, collaborative filtering, deep learning, and other technologies is designed and implemented. This algorithm comprehensively considers various features of the asset profile and investor profile to calculate the matching degree. When generating the recommendation list, the algorithm prioritizes selecting assets whose risk levels match the investor's risk preferences, and, under the same matching degree, prioritizes recommending assets with lower risk. Assets with low risk or low disposal difficulty; two-way feedback mechanism and model optimization: collect investor behavior data such as clicks, favorites, purchase requests, and transactions of recommended assets, as well as explicit feedback such as satisfaction ratings of recommendation results and reasons for disinterest; collect transaction result data such as actual transaction cycle, transaction price, disposal success rate, and subsequent disposal status of assets as implicit feedback; input explicit and implicit feedback data into the recommendation algorithm model, and dynamically adjust the parameters of the asset risk assessment model, investor risk preference model, and intelligent matching algorithm through mechanisms such as reinforcement learning or online learning to achieve continuous optimization and self-adaptation of the recommendation system; Explanation of Recommendation Results and Risk Warnings: The recommendation results provide an explanation of the asset's risk level, main risk points, and the reasons for the recommendation. For high-risk assets, clear risk warnings and necessary due diligence suggestions are provided.

2. A two-way feedback intelligent recommendation system for special assets that integrates risk assessment factors, characterized in that: include: The profile building module is used to create profiles of specific assets and investors; The risk fusion module is used to match and integrate asset risk factors with investors' risk preferences; The intelligent recommendation module is used to execute intelligent matching algorithms and generate personalized recommendation lists; The feedback collection module is used to collect explicit feedback from investors and implicit feedback on transaction results; The model optimization module is used to iteratively optimize the recommendation algorithm model based on feedback data; the explanation and prompt module is used to provide explanations of the recommendation results and risk prompts.

3. The method according to claim 1, characterized in that, The asset profile includes asset type, geographical location, value range, ownership status, mortgage status, litigation status, disposal difficulty, potential returns, risk score, and risk label.

4. The method according to claim 1, characterized in that, Investor profiles include browsing behavior, search history, collection preferences, transaction records, financial strength, risk appetite, regional restrictions, disposal experience, and risk tolerance.

5. The method according to claim 1, characterized in that, The intelligent matching algorithm integrates multiple technologies such as content matching, collaborative filtering, and deep learning.

6. The method according to claim 1, characterized in that, Two-way feedback includes explicit feedback from investors and implicit feedback on transaction results.

7. The method according to claim 1, characterized in that, The model parameters are dynamically adjusted through reinforcement learning or online learning mechanisms.

Images