Hybrid recommender system based on fuzzy neural algorithm

The internet has provided much information to the users, and the huge amount of information has caused the overhead information problem so that recommender systems, as a solution by recommending related items, try to find the users favorite items. Most recommender systems recommend specific items based on user profile settings and past privileges. Various approaches, such as collaborative refinement and content-based refinement, are used to build the recommender system. The proposed framework utilizes popular and unpopular user-specific items and is built on the integration of the fuzzy neural network associative exploration approach (Anfis) with a collaborative refinement approach. This study proposes a hybrid recommender system that provides two-dimensional space (user * Item) ancillary information to reflect users' preferences and items. Experimental results from experiments on the actual Movielens 1 m 6040 dataset and 3952 videos show that the mean error (MAE) of 2.79 compared to other low-resolution methods also predicts a correct percentage of 21.75 to other windshields. Besides, experimental results show higher precision of the proposed system comparing to the former recommender systems.