WCSE

Selecting Classification Model for the Personalized Movie Recommendation System by Feature Adjustment Method

Abstract— Recommendation systems are widely used to improve market potential in theater business today. However, the efficiency of the personalized movie recommendation system (PMRS) using model-based techniques is related to employed classifier and a number of features. This research aims to select a suitable classification model by feature adjustment method for creating the recommendation rules of PMRS. The suggestion model is appraised using retrieval performance measures by Accuracy between 3 algorithms of classification consisting of J48, Naïve Bayes (NB) and Multilayer Perceptron (MLP). The datasets for model construction are collected through surveying from 383 movie audiences who live in Nakhon-Ratchasima province, Thailand. The results of the accuracy performance show that J48 algorithm produces the finest accuracy (70.28%) followed by NB (68.28%) and MLP (66.23%), respectively. In addition, the performance of J48 by feature adjustment method provides 58 combinations which are created from 6 features of movie audience’s profile and 19 features of movie genres. The results of feature adjustment method present the consistency between accuracy performance and a number of features. However, the progress of recommendation rules set selection for PMRS will be chosen only 36 high performance combinations of adjustment features and these combinations will be applied to the development of a new personalized movie recommendation system.

Index Terms— personalized recommendation, classification, feature adjustment.

Supachanun Wanapu
Faculty of Management Science, Nakhon Ratchasima Rajabhat University, THAILAND
Thawatphong Phithak
School of Information Technology, Suranaree University of Technology, THAILAND
Narodom Kittidachanupap
Faculty of Science Technology and Agriculture, Yala Rajabhat University, THAILAND

[Download]

Cite: Supachanun Wanapu, Thawatphong Phithak, Narodom Kittidachanupap, "Selecting Classification Model for the Personalized Movie Recommendation System by Feature Adjustment Method," Proceedings of 2016 6th International Workshop on Computer Science and Engineering, pp. 682-686, Tokyo, 17-19 June, 2016.