Personalized recommendation engine system and method based on multiple Agents

Abstract

The invention relates to the technical field of recommendation engines, in particular to a personalized recommendation engine system and method based on multiple Agents. According to the technical scheme, the personalized recommendation engine system and method based on the multiple Agents comprise a content analysis module; compared with a traditional multi-Agent-based personalized recommendation engine system, the multi-Agent-based personalized recommendation engine system has the advantages that a relatively fixed system architecture is often adopted, the coupling degree between modules is relatively high, the flexibility and the expandability of the system are limited, meanwhile, a used personalized recommendation algorithm may be relatively single, or although multiple algorithms are used, an effective fusion mechanism is lacked, and the system is relatively low in cost. According to the personalized recommendation engine system based on the multiple Agents, a modular design idea is adopted, so that each Agent can independently complete a specific task and can cooperatively work with other Agents, meanwhile, the coupling degree between modules is reduced, the flexibility and expandability of the system are improved, and the user experience is improved. Therefore, the system can be customized and optimized according to different requirements and scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of recommendation engines, and in particular to a multi-agent-based personalized recommendation engine system and method. Background Art

[0002] The personalized recommendation engine system based on multi-agent is an innovative intelligent recommendation system, which uses multiple agents to work together to provide users with personalized recommendations more accurately and efficiently.

[0003] Existing multi-agent based personalized recommendation engine systems often adopt a relatively fixed system architecture with a high degree of coupling between modules, which limits the flexibility and scalability of the system. At the same time, the personalized recommendation algorithms used may be relatively single, or although multiple algorithms are used, there is a lack of effective fusion mechanism.

[0004] The existing multi-agent-based personalized recommendation engine system often adopts a relatively fixed system architecture, and the coupling degree between modules is high, which leads to the problem that the flexibility and scalability of the system are limited. This multi-agent-based personalized recommendation engine system adopts a modular design concept, so that each agent can complete specific tasks independently and can work collaboratively with other agents. At the same time, the coupling degree between modules is reduced, which improves the flexibility and scalability of the system, and enables the system to be customized and optimized according to different needs and scenarios. Summary of the invention

[0005] In order to overcome the problem that the existing multi-agent based personalized recommendation engine systems often adopt a relatively fixed system architecture, the coupling degree between modules is high, resulting in limited flexibility and scalability of the system. At the same time, the personalized recommendation algorithm used may be relatively single, or although multiple algorithms are used, there is a lack of effective fusion mechanism.

[0006] The technical solution of the present invention is: a personalized recommendation engine system based on multi-agent, comprising:

[0007] Data collection module: used to collect user data from various data sources;

[0008] User portrait module: used to receive data collected by the data collection module to update the user portrait;

[0009] Content analysis module: used to conduct in-depth analysis of recommended content;

[0010] Inference algorithm module: used to perform recommendation calculations based on the user portrait module;

[0011] Context-aware module: used to monitor and analyze user context information in real time.

[0012] Preferably, the data collection module is responsible for collecting user data from various data sources, including but not limited to user behavior logs, social media, e-commerce websites, etc., while cleaning and preprocessing the data to ensure the accuracy and consistency of the data, extract key features, and provide a basis for subsequent analysis.

[0013] Preferably, the user portrait module is used to build and update user portraits based on the collected user data, analyze the user's interests, preferences, and purchasing habits, form a personalized user model, and identify the user's potential needs and interests to provide a basis for recommendations.

[0014] Preferably, the content analysis module is used to conduct in-depth analysis and understanding of the recommended content, extract the content's features, tags, and metadata for matching with the user portrait, and evaluate the quality, relevance, and popularity of the content to ensure the accuracy of the recommendation.

[0015] Preferably, the recommendation algorithm module is used to apply various recommendation algorithms, including but not limited to collaborative filtering, content recommendation, and hybrid recommendation, to perform recommendation calculations, generate personalized recommendation lists based on user portraits and content analysis results, and continuously optimize and adjust algorithm parameters to improve the accuracy and diversity of recommendations.

[0016] Preferably, the context-aware module is used to monitor and analyze the user's context information in real time, including but not limited to time, location, device, network environment, etc., and adjust the recommendation strategy based on the context information to provide recommendations that better meet the user's current needs, predict the user's future behavior and preferences, and make personalized recommendations in advance.

[0017] Preferably, a modular design is adopted so that each agent can complete specific tasks independently and can work in collaboration with other agents, which improves the flexibility and scalability of the system and enables the system to be customized and optimized according to different needs and scenarios. At the same time, multiple agents exchange information and share resources through intelligent collaborative algorithms, thereby improving the accuracy and diversity of recommendations.

[0018] A personalized recommendation method based on multi-agent includes the following steps:

[0019] S11: First, construct a user portrait and update the user portrait in real time;

[0020] S12: Perform in-depth analysis and understanding of the recommended content according to the content analysis module, and extract content features, tags, and metadata;

[0021] S13: Optimize based on user feedback and the calculation results of the recommendation algorithm module.

[0022] Preferably, when constructing and updating the user profile, the following methods are included:

[0023] S21: using a data collection agent to collect user data from various data sources, including but not limited to user behavior logs, social media, e-commerce websites, etc.;

[0024] S22: Clean and preprocess data to ensure data accuracy and consistency;

[0025] S23: extract key features, including but not limited to user interests, preferences, and purchasing habits;

[0026] S24: Based on the extracted features, construct a user portrait to form a personalized user model;

[0027] S25: As user data continues to accumulate, user portraits are updated regularly to maintain their accuracy and timeliness.

[0028] Preferably, when performing multi-agent collaborative recommendation, the following method is included:

[0029] S31: Content analysis agent conducts in-depth analysis and understanding of recommended content, extracting content features, tags, and metadata;

[0030] S32: The recommendation algorithm Agent applies various recommendation algorithms, including but not limited to collaborative filtering, content recommendation, and hybrid recommendation, to perform recommendation calculations. At this time, it will combine user portraits and content analysis results to generate a preliminary personalized recommendation list;

[0031] S33: The context-aware Agent monitors and analyzes the user's context information in real time, including but not limited to time, location, device, and network environment, and adjusts the recommendation strategy based on this information to provide recommendations that better meet the user's current needs;

[0032] S34: Multiple agents work together to optimize the recommendation results.

[0033] Preferably, when performing user feedback and recommendation optimization, the following methods are included:

[0034] S41: using the feedback processing Agent to collect user feedback on the recommended content;

[0035] S42: Analyze feedback data and evaluate recommendation effects and user satisfaction, including but not limited to statistical analysis of user behavior, evaluation analysis of recommended content, and survey of user satisfaction;

[0036] S43: Optimizing the recommendation model according to the feedback analysis results;

[0037] S44: Reapply the optimized recommendation model to the recommendation process, and continue to collect and analyze user feedback to continuously iterate and optimize the recommendation system.

[0038] Beneficial effects of the present invention:

[0039] 1. Compared with the traditional personalized recommendation engine system based on multi-agent, which often adopts a relatively fixed system architecture, the coupling degree between modules is high, resulting in limited flexibility and scalability of the system. At the same time, the personalized recommendation algorithm used may be relatively single, or although multiple algorithms are used, there is a lack of effective fusion mechanism. The personalized recommendation engine system based on multi-agent adopts a modular design concept, so that each agent can complete specific tasks independently and can work with other agents at the same time. At the same time, the coupling degree between modules is reduced, which improves the flexibility and scalability of the system, and enables the system to be customized and optimized according to different needs and scenarios;

[0040] 2. The modular design enables each agent to complete specific tasks independently and work collaboratively with other agents, which improves the flexibility and scalability of the system and enables the system to be customized and optimized according to different needs and scenarios. At the same time, multiple agents exchange information and share resources through intelligent collaborative algorithms, thereby improving the accuracy and diversity of recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 What is shown is a flowchart of a personalized recommendation method based on multi-agent of the present invention;

[0042] Figure 2 What is shown is a schematic diagram of a user portrait construction and update process of a personalized recommendation method based on multi-agents of the present invention;

[0043] Figure 3 What is shown is a schematic diagram of a multi-agent collaborative recommendation process of a multi-agent-based personalized recommendation method of the present invention;

[0044] Figure 4 What is shown is a schematic diagram of a user feedback and recommendation optimization process of a multi-agent-based personalized recommendation method of the present invention. DETAILED DESCRIPTION

[0045] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] See also Figure 1-4The present invention provides an embodiment: a personalized recommendation engine system based on multi-agent, comprising:

[0047] Data collection module: used to collect user data from various data sources;

[0048] User portrait module: used to receive data collected by the data collection module to update the user portrait;

[0049] Content analysis module: used to conduct in-depth analysis of recommended content;

[0050] Inference algorithm module: used to perform recommendation calculations based on the user portrait module;

[0051] Context-aware module: used to monitor and analyze user context information in real time.

[0052] Preferably, the data collection module is responsible for collecting user data from various data sources, including but not limited to user behavior logs, social media, e-commerce websites, etc., while cleaning and preprocessing the data to ensure the accuracy and consistency of the data, extract key features, and provide a basis for subsequent analysis.

[0053] Preferably, the user portrait module is used to build and update user portraits based on the collected user data, analyze the user's interests, preferences, and purchasing habits, form a personalized user model, and identify the user's potential needs and interests to provide a basis for recommendations.

[0054] Preferably, the content analysis module is used to conduct in-depth analysis and understanding of the recommended content, extract the content's features, tags, and metadata for matching with the user portrait, and evaluate the content's quality, relevance, and popularity to ensure the accuracy of the recommendations.

[0055] Preferably, the recommendation algorithm module is used to apply various recommendation algorithms, including but not limited to collaborative filtering, content recommendation, and hybrid recommendation, to perform recommendation calculations, generate personalized recommendation lists based on user portraits and content analysis results, and continuously optimize and adjust algorithm parameters to improve the accuracy and diversity of recommendations.

[0056] Preferably, the context-aware module is used to monitor and analyze the user's context information in real time, including but not limited to time, location, device, network environment, etc., and adjust the recommendation strategy based on the context information to provide recommendations that better meet the user's current needs, predict the user's future behavior and preferences, and make personalized recommendations in advance.

[0057] Preferably, a modular design is adopted so that each agent can complete specific tasks independently and can work in collaboration with other agents, which improves the flexibility and scalability of the system and enables the system to be customized and optimized according to different needs and scenarios. At the same time, multiple agents exchange information and share resources through intelligent collaborative algorithms, thereby improving the accuracy and diversity of recommendations.

[0058] A personalized recommendation method based on multi-agent includes the following steps:

[0059] S11: First, construct a user portrait and update the user portrait in real time;

[0060] S12: Perform in-depth analysis and understanding of the recommended content according to the content analysis module, and extract content features, tags, and metadata;

[0061] S13: Optimize based on user feedback and the calculation results of the recommendation algorithm module.

[0062] Preferably, when constructing and updating the user profile, the following methods are included:

[0063] S21: using a data collection agent to collect user data from various data sources, including but not limited to user behavior logs, social media, e-commerce websites, etc.;

[0064] S22: Clean and preprocess data to ensure data accuracy and consistency;

[0065] S23: extract key features, including but not limited to user interests, preferences, and purchasing habits;

[0066] S24: Based on the extracted features, construct a user portrait to form a personalized user model;

[0067] S25: As user data continues to accumulate, user portraits are updated regularly to maintain their accuracy and timeliness.

[0068] Preferably, when performing multi-agent collaborative recommendation, the following method is included:

[0069] S31: Content analysis agent conducts in-depth analysis and understanding of recommended content, extracting content features, tags, and metadata;

[0070] S32: The recommendation algorithm agent applies various recommendation algorithms, including but not limited to collaborative filtering, content recommendation, and hybrid recommendation, to perform recommendation calculations. At this time, it will combine user portraits and content analysis results to generate a preliminary personalized recommendation list;

[0071] S33: The context-aware Agent monitors and analyzes the user's context information in real time, including but not limited to time, location, device, and network environment, and adjusts the recommendation strategy based on this information to provide recommendations that better meet the user's current needs;

[0072] S34: Multiple agents work together to optimize the recommendation results.

[0073] Preferably, when performing user feedback and recommendation optimization, the following methods are included:

[0074] S41: using the feedback processing Agent to collect user feedback on the recommended content;

[0075] S42: Analyze feedback data and evaluate recommendation effects and user satisfaction, including but not limited to statistical analysis of user behavior, evaluation analysis of recommended content, and survey of user satisfaction;

[0076] S43: Optimizing the recommendation model according to the feedback analysis results;

[0077] S44: Reapply the optimized recommendation model to the recommendation process, and continue to collect and analyze user feedback to continuously iterate and optimize the recommendation system.

[0078] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of those skilled in the art without departing from the spirit of the present invention.

Claims

1. A personalized recommendation engine system based on multi-agent; characterized by: Included are: Data collection module: used to collect user data from various data sources; User portrait module: used to receive data collected by the data collection module to update the user portrait; Content analysis module: used to conduct in-depth analysis of recommended content; Inference algorithm module: used to perform recommendation calculations based on the user portrait module; Context-aware module: used to monitor and analyze user context information in real time.

2. The multi-agent-based personalized recommendation engine system according to claim 1, characterized in that: The data collection module is responsible for collecting user data from various data sources, including but not limited to user behavior logs, social media, e-commerce websites, etc., and cleaning and preprocessing the data to ensure the accuracy and consistency of the data, extract key features, and provide a basis for subsequent analysis.

3. The multi-agent-based personalized recommendation engine system according to claim 2, characterized in that: The user portrait module is used to build and update user portraits based on the collected user data, analyze user interests, preferences, and purchasing habits, form a personalized user model, and identify users' potential needs and interests to provide a basis for recommendations.

4. The multi-agent-based personalized recommendation engine system according to claim 3, characterized in that: The content analysis module is used to conduct in-depth analysis and understanding of recommended content, extract content features, tags, and metadata to match them with user portraits, and evaluate the quality, relevance, and popularity of content to ensure the accuracy of recommendations.

5. The multi-agent-based personalized recommendation engine system according to claim 4, characterized in that: The recommendation algorithm module is used to apply various recommendation algorithms, including but not limited to collaborative filtering, content recommendation, and hybrid recommendation, to perform recommendation calculations, generate personalized recommendation lists based on user portraits and content analysis results, and continuously optimize and adjust algorithm parameters to improve the accuracy and diversity of recommendations.

6. The multi-agent-based personalized recommendation engine system according to claim 5, characterized in that: The context-aware module is used to monitor and analyze the user's context information in real time, including but not limited to time, location, device, network environment, etc., and adjust the recommendation strategy based on the context information to provide recommendations that better meet the user's current needs, predict the user's future behavior and preferences, and make personalized recommendations in advance.

7. A multi-agent-based personalized recommendation method, according to the multi-agent-based personalized recommendation engine system of claim 1, characterized in that: The following steps are included: S11: First, construct a user portrait and update the user portrait in real time; S12: Perform in-depth analysis and understanding of the recommended content according to the content analysis module, and extract content features, tags, and metadata; S13: Optimize based on user feedback and the calculation results of the recommendation algorithm module.

8. The multi-agent-based personalized recommendation method according to claim 7, characterized in that: When building and updating user portraits, the following methods are included: S21: using a data collection agent to collect user data from various data sources, including but not limited to user behavior logs, social media, e-commerce websites, etc.; S22: Clean and preprocess data to ensure data accuracy and consistency; S23: extract key features, including but not limited to user interests, preferences, and purchasing habits; S24: Based on the extracted features, construct a user portrait to form a personalized user model; S25: As user data continues to accumulate, user portraits are updated regularly to maintain their accuracy and timeliness.

9. The multi-agent-based personalized recommendation method according to claim 8, characterized in that: When performing multi-agent collaborative recommendation, the following methods are included: S31: Content analysis agent conducts in-depth analysis and understanding of recommended content, extracting content features, tags, and metadata; S32: The recommendation algorithm agent applies various recommendation algorithms, including but not limited to collaborative filtering, content recommendation, and hybrid recommendation, to perform recommendation calculations. At this time, it will combine user portraits and content analysis results to generate a preliminary personalized recommendation list; S33: The context-aware Agent monitors and analyzes the user's context information in real time, including but not limited to time, location, device, and network environment, and adjusts the recommendation strategy based on this information to provide recommendations that better meet the user's current needs; S34: Multiple agents work together to optimize the recommendation results.

10. The multi-agent-based personalized recommendation method according to claim 9, characterized in that: When optimizing user feedback and recommendations, the following methods are included: S41: using the feedback processing Agent to collect user feedback on the recommended content; S42: Analyze feedback data and evaluate recommendation effects and user satisfaction, including but not limited to statistical analysis of user behavior, evaluation analysis of recommended content, and survey of user satisfaction; S43: Optimizing the recommendation model according to the feedback analysis results; S44: Reapply the optimized recommendation model to the recommendation process, and continue to collect and analyze user feedback to continuously iterate and optimize the recommendation system.

Images