Dynamic Information Recommendation System for Viewing Fatigue
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information recommendation systems primarily focus on recommending favorite information to users, leading to viewing fatigue as the same information is repeatedly presented, without incorporating random recommendations or updating based on re-recorded user behavior.
Innovation Solution
A method and system that involve a terminal device and server, where the server recommends first, second, and third types of information based on user behavior records, with the option to display a random-recommendation button to trigger random recommendations, and adjust recommendations based on user interaction and behavior records over distinct time periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the system recommends favorite information based on user behavior records, then the recommendation accuracy is improved, but viewing fatigue occurs due to repeated presentation of the same information
Solution Approach 1:
The recommendation system dynamically adjusts between two modes: a first time period where recommendations are based on user behavior records (static personalization), and a second time period where random recommendations are provided (dynamic variation). This temporal dynamic resolves the contradiction by alternating between accuracy-focused and variety-focused recommendation strategies, preventing viewing fatigue while maintaining personalization benefits.
Solution Approach 2:
The system implements periodic action by dividing recommendation delivery into distinct time periods with different strategies. The first time period uses behavior-based recommendations, the second time period introduces random recommendations to break monotony, and the third time period returns to behavior-based recommendations with updated behavior records. This periodic alternation prevents continuous exposure to the same content types, alleviating viewing fatigue while preserving recommendation accuracy.
2Adaptability or versatility
If the system provides random recommendations, then viewing variety is improved, but recommendation accuracy decreases compared to behavior-based recommendations
Solution Approach 1:
The recommendation system segments the recommendation process into distinct time periods with different strategies. The first time period focuses on accuracy with behavior-based recommendations, the second time period focuses on variety with random recommendations, and the third time period combines both approaches. This segmentation allows each mode to excel at its intended function without compromising the other, as they operate in temporally separated phases.
Solution Approach 2:
The system dynamically switches between accuracy-oriented and variety-oriented recommendation modes based on the current time period. During the first time period, the system prioritizes accuracy through behavior analysis. During the second time period, it prioritizes variety through random selection. This dynamic switching resolves the contradiction by allowing both accuracy and variety to take turns being the primary objective, rather than constantly competing.
3Adaptability or versatility
If the system continuously updates behavior records and recommendations, then personalization is improved, but system complexity increases
Solution Approach 1:
The system uses periodic action to manage complexity by updating behavior records and switching recommendation strategies at defined time period boundaries rather than continuously. The first behavior record is collected during the first time period, the second behavior record is collected during the second time period, and recommendations are updated at these discrete intervals. This periodic update mechanism maintains personalization effectiveness while avoiding the complexity of continuous real-time updates.
Data Source
AI summary
An information recommendation method and a related device. The method includes: receiving first recommended information recommended by a server in a first time period, the first recommended information being information associated with a first type of information associated with viewed information of a user with a viewed amount not less than a first threshold and determined according to a first behavior record of the user; receiving second recommended information recommended by the server in a second time period; and receiving third recommended information recommended by the server in a third time period, the third recommended information being information associated with a second type of information associated with second viewed information of the user with a second viewed amount not less than a second threshold and determined according to a second behavior record of the user.