Recommendation Systems Based on Online User's Action

In this paper, we propose a new recommender algorithm based on multi-dimensional users behavior and new measurements. It's used in the framework of our recommender system that use knowledge discovery techniques to the problem of making product recommendations during a live user interaction. Most of Collaborative filtering algorithms based on user's rating or similar item that other users bought, we propose to combine all user's action to predict recommendation. These systems are achieving widespread success in E-tourism nowadays. We evaluate our algorithm on tourism dataset. Evaluations have shown good results. We compared our algorithm to Slope One and Weight Slope One. We obtained an improvement of 5% in precision and recall. And an improvement of 12% in RMSE and nDCG.