Abstract

Recommender systems are those systems which try to provide some suggestions that have accordance with personal desires and help him/her in the decision making process, based on performance, personal preferences, Behavior of user and depending on the context in which they are used. For this purpose, several algorithms have been used that most of them are based on two algorithms, Collaborative based Filtering and content based filtering. In this study, one method is presented which will offer recommendations based on users’ comments. For this purpose, a data set is formed of the comments about movies, then some features are extracted from comments of users by using Ontology and text processing, and then provides recommendations by using K algorithm of nearest neighbor (KNN) and based on user profiling and profiling items. Obtained results of this method show that effects of user’s comments on the accuracy of recommender systems are appropriate and by considering that this method can also be used to improve the content based methods, it will be very useful. One of the other important benefits versus other methods is privacy.Keywords: Recommender system, opinion mining, Ontology, K algorithm of nearest neighbor, item/user’s profile.