GroupLens: an open architecture for collaborative filtering of netnews
CSCW '94 Proceedings of the 1994 ACM conference on Computer supported cooperative work
A language modeling approach to information retrieval
Proceedings of the 21st annual international ACM SIGIR conference on Research and development in information retrieval
Authoritative sources in a hyperlinked environment
Proceedings of the ninth annual ACM-SIAM symposium on Discrete algorithms
Item-based collaborative filtering recommendation algorithms
Proceedings of the 10th international conference on World Wide Web
Methods and metrics for cold-start recommendations
SIGIR '02 Proceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval
A Linguistically Motivated Probabilistic Model of Information Retrieval
ECDL '98 Proceedings of the Second European Conference on Research and Advanced Technology for Digital Libraries
A generalized Co-HITS algorithm and its application to bipartite graphs
Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining
| Hi-index | 0.00 |
Recently, there has been a growing need for more sophisticated recommendation techniques with an increase in the amount of data available on the Web. In this study, we especially focus on recommending items with long text, and aim at achieving this using a method of link analysis of a user-item bipartite graph in a regularization framework based on Co-HITS algorithm. This method can integrate, via mutual reinforcement, the graph structure and the content of both user profiles and items. It has never been seen in the mainstream of conventional recommendation techniques. In our experiments, we used the data of Web browsing history, assuming Web news articles as target items. We evaluated the list of top-N items recommended based on the browsing history, using a test set that consists of a part of viewed items for each user. We demonstrate through the experiments that the proposed method outperformed several baseline methods in a situation where only a small amount of browsing behavior is observed.