Article Recommendation System Using Inverted Attention Degrees
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current article recommendation systems lack diversity as they primarily suggest articles with similar content based on users' historical reads, leading to a limited range of recommendations.
Innovation Solution
An article recommendation method that analyzes attention degrees between historical articles and candidate reference articles to identify articles with lower recommendation potential, using these as references to suggest diverse articles by leveraging the historical reading patterns of users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If articles with similar content are recommended to users based on historical reading patterns, then recommendation accuracy is improved, but recommendation diversity deteriorates
Solution Approach 1:
Instead of selecting articles with high attention degrees (similar content) for recommendation, the patent inverts the approach by selecting articles with low attention degrees from historical data as reference articles. This inversion enables the system to recommend diverse content that users have rarely engaged with, thereby improving recommendation diversity while maintaining accuracy through the attention mechanism
Solution Approach 2:
The patent changes the selection parameter from attention degree (high value) to attention degree (low value) when choosing reference articles. By flipping the selection criterion from maximizing to minimizing attention degree, the system identifies articles that are less likely to be recommended traditionally, thus expanding diversity while preserving personalization through the attention-based filtering mechanism
Data Source
AI summary
This application provides an article recommendation method and apparatus, a computer device, and a storage medium. The method includes: obtaining a historical reading article set of a to-be-analyzed target user, the historical reading article set comprising a plurality of historical articles that have been read by the target user at different reading moments; determining, for each historical article of the plurality of historical articles, first attention degree values between the each historical article and candidate reference historical articles having reading moments before that of the each historical article, the first attention degree values reflecting a possibility that a user is recommended to read the historical article if the user has read the candidate reference historical articles; selecting at least one recommendation reference article from the candidate reference historical articles corresponding to the historical articles according to the first attention degree values between the historical articles and corresponding candidate reference historical articles; and determining, in a recommendable article set at least one candidate recommended article to be recommended to the target user based on the at least one recommendation reference article.