Improving Recommender System Based on Item's Structural Information in Affinity Network

Ma’Arif, Muhammad Rifqi • Mulyanto, Agus

Abstract

This paper proposes a technique to improve the accuracy of recommender system result which employ collaborative filtering technique. The proposed method incorporates structural equivalence score of items in affinity network into collaborative filtering technique. Structural equivalence is one of important concept in social network analysis which captures the similarity of items regarding their structural position on the affinity network. Nowadays, various concepts within social network analysis are widely use in many domains to provide better analytical framework. In this paper, we will use structural equivalence of items to enhance the calculation of items similarity as a part of collaborative filtering method. We tested our approach on Netflix database. Then, based on our results we can conclude that considering the structural information of item in affinity network is indeed beneficial.

Metrics

  • 3 views
  • 0 downloads

Conference

1st International Conference on Electrical Engineering, Computer Science and Informatics 2014

The 1st International Conference on Electrical Engineering Computer Science and Informatics (EECS... see more