Interactive Recommendation System Balancing Precision and Diversity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing recommendation systems face challenges in balancing precision and diversity, leading to user fatigue due to the inability to utilize user interaction behavior data in real-time, resulting in weak user interaction perception and monotonous information flow.
Innovation Solution
An interactive recommendation method that records user behavior and sends product information to a server for sorting and filtering similar products, inserting a second area on the page for displaying these recommendations, which adapts in size based on user scrolling actions and incorporates diversity strategies to down-weight categories, ensuring both precision and diversity in recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If paginated recommendation approach is adopted, then product information can be organized across multiple pages, but user behavior data from previous pages cannot be collected and utilized in a timely manner
Solution Approach 1:
The patent implements real-time feedback mechanisms by capturing user behavior data (clicks, views,停留 time) on the current page and immediately using this feedback to generate and display personalized recommendations. The system continuously monitors user interactions and adjusts recommendation content dynamically, ensuring that user behavior data is collected and utilized without time delay, thus resolving the information loss and time delay issues inherent in traditional paginated approaches.
Solution Approach 2:
The patent maintains continuous useful action by seamlessly integrating recommendation content within the same page view, allowing user behavior data collection and recommendation generation to occur continuously without page transitions. The system processes user interactions in real-time and continuously updates recommendation content, eliminating the interruptions caused by pagination and ensuring uninterrupted data flow and utilization.
2Measurement precision
If overly precise recommendations are provided, then recommendation precision is improved, but diversity is reduced leading to user fatigue
Solution Approach 1:
The patent applies local quality by differentiating the characteristics of recommendation content displayed in different areas of the page. The first area (original recommendation content) maintains high precision based on user profile and historical behavior, while the second area (inserted recommendation content) emphasizes diversity by incorporating trending products, promotional items, and category-expanding suggestions. This spatial differentiation of recommendation qualities allows the system to simultaneously satisfy precision and diversity requirements without causing user fatigue.
Solution Approach 2:
The patent implements dynamics by making the recommendation system adaptable and flexible in response to real-time user interactions. The system dynamically adjusts the mix of precise and diverse recommendations based on user behavior patterns, session context, and engagement metrics. When users show signs of fatigue with precise recommendations, the system automatically increases diversity in subsequent recommendations, creating a dynamic balance that prevents user fatigue while maintaining relevance.
3Ease of operation
If a second area is inserted on the page to display similar products, then real-time recommendations based on user interaction are achieved, but page layout complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the page into distinct functional areas: the first area for original recommendation content and the second area for inserted similar product recommendations. Each area serves a specific purpose and can be independently managed, styled, and optimized. This segmentation allows the system to implement complex recommendation logic while maintaining a clean, organized page layout that is easy to navigate and understand, thus resolving the contradiction between enhanced interactivity and layout complexity.
Data Source
AI summary
This application provides an interactive recommendation method, electronic device, and storage medium. The method comprises displaying recommended product information in a first area of a webpage; in response to a user's trigger action on a current product in the first area, recording the user's behavior into feature information, which includes user characteristics and product characteristics; sending the current product's information and feature information to a server; receiving information of similar products returned from the server, wherein the similar products are selected by the server from a product database and resulted from sorting and filtering based on the feature information; inserting a second area of original size on the webpage; and displaying information of the similar products in the second area. According to the embodiments of this application, recommendations are made based on user interaction behavior, achieving a balance between precision and diversity in recommendations.