Recommender systems aim to automatically provide users with personalized information in an overloaded search space. To dual with vagueness and imprecision problems in RS, several researches have proposed fuzzy-based approaches. Even though these works have incorporated experimental evaluation, they were used in different recommendation scenarios which makes it difficult to have a fair comparison between them. Also, some of them performed an items and/or users clustering before generating recommendations. For this reason, they need additional information such as item attributes or trust between users which are not always available. In this paper, the authors propose to use fuzzy set techniques to predict the rating of a target user for each unrated item. It uses the target user's history in addition with rating of similar users which allows to the target user to contribute in the recommendation process. Experimental results on several datasets seem to be promising in term of MAE (mean average error), RMSE (root mean square error), accuracy, precision, recall, and f-measure.
展开▼
