An Improved Collaborative Filtering Recommendation Algorithm Based on Retroactive Inhibition Theory

Collaborative filtering (CF) is the most classical and widely used recommendation algorithm, which mainly to predict user preferences by mining user’s historical data. CF algorithms can be divided into two main categories: user-based item-based CF, recommend items based on rating information from similar profiles (user-based) or similarity between (item-based). However, since are not static, it vital take account changing of users when making recommendations achieve more accurate recommendations. In recent years, there have been studies using memory as a factor measure changes in preference exploring retention relationship forgetting mechanism time. Nevertheless, according theory inhibition, factors that cause retroactive inhibition proactive mere evolutions over Therefore, our work proposed method combines traditional algorithm (namely, RICF) accurately explore evolution preferences. Meanwhile, embedding training introduced represent features better alleviate problem data sparsity, then item embeddings clustered points different items. Moreover, we conducted experiments real-world datasets demonstrate practicability RICF. The show RICF performs interpretable than collaborative well state-of-art sequential models such LSTM GRU.