Hybrid System Based on Intelligent Neighbor Formation Algorithm

Currently, people prefer shopping interesting items from the Internet. Therefore, e-business had installed recommender systems to support users in selecting products as quick as possible. The responsibility of the recommender system is to propose interesting items to the target users in order to gain their interests. Unfortunately, the existing recommender systems have some imperfect function under some certain situations. This paper proposed the INFA that improves the qualities of neighbors by detailing features of users´ preferences. Moreover, in the neighbor forming process, the consideration domain of agreements is classified into two independent domains: the positive, and the negative agreements. Furthermore, the MNBT has improved the ability of recommender under the situation of no opinion from neighbors, or no existing neighbors