User-Oriented Preference Toward a Recommender System

Abstract

Nowadays, it is convenient for us to use a search engine to get our needed information. But sometimes it will misunderstand the information because of the different media reports. The Recommender System (RS) is popular to use for every business since it can provide information for users that will attract more revenues for companies. But also, sometimes the system will recommend unneeded information for users. Because of this, this paper provided an architecture of a recommender system that could base on useroriented preference. This system is called UOP-RS. To make the UOP-RS significantly, this paper focused on movie theatre information and collect the movie database from the IMDb website that provides information related to movies, television programs, home videos, video games, and streaming content that also collects many ratings and reviews from users. This paper also analyzed individual user data to extract the user's features. Based on user characteristics, movie ratings/scores, and movie results, a UOP-RS model was built. In our experiment, 5000 IMDb movie datasets were used and 5 recommended movies for users. The results show that the system could return results on 3.86 s and has a 14% error on recommended goods when training data as = 50. At the end of this paper concluded that the system could quickly recommend users of the goods which they needed. The proposed system will extend to connect with the Chatbot system that users can make queries faster and easier from their phones in the future.