원문정보
초록
영어
The tremendous growth of data in recent years poses some key challenges for recommender systems. Theses keys are related with producing high quality recommendations and fast performing the composition recommended items. In this paper, we propose social clustering-based similar user index to not only improve the prediction of recommendations, but also compose personalized recommendations in fast. Through the experimental result, we show that proposed clustering method is more accurate than k-means which is prevalent clustering techniques. And, we reduce computation time needed for composing recommendation. That is, proposed clustering-based indexing method improves the performance of recommender systems which deals with a very large data.
목차
1. Introduction
2. Related Works and Backgrounds
2.1. Collaborative Filtering
2.2. Clustering
2.3. Recommender Algorithms with the Social Network
3. Social Clustering-based Similar User Indexing Mechanism
3.1. Generating Social Network with the Recommender Dataset
3.2. Three Steps-clustering for Composing the Similar User Index
3.3. Searching Similar Users with the Similar User Index
4. Evaluation
4.1. Response Time
4.2. Clustering Accuracy
5. Conclusion
Acknowledgement
References
키워드
저자정보
참고문헌
- 1(Reference title not available)
- 2(Reference title not available)
- 3Empirical analysis of predictive algorithms for collaborative filtering네이버 원문 이동
- 4(Reference title not available)
- 5Discovering shared interests using graph analysis네이버 원문 이동
- 6A K‐Means Clustering Algorithm네이버 원문 이동
- 7(Reference title not available)
- 8CURE: an efficient clustering algorithm for large databases네이버 원문 이동
- 9Recommender systems survey네이버 원문 이동
- 10Algorithms for clustering data네이버 원문 이동
- 11Efficient Recommender System based on Graph Data for Multimedia Application네이버 원문 이동