Paper recommendation; data mining; singular value decomposition
Issue Date
2015-09
Publisher
ACM Hypertext
Citation
HT '15 Proceedings of the 26th ACM Conference on Hypertext & Social Media, Page. 329-330
Abstract
To recommend newly-published papers that did not receive any citations yet, this paper proposes a novel method by using the authors’ interest on the papers cited in the newlypublished paper. Compared to citation-network based methods, the empirical validation shows a significant improvement with our method in accuracy of 11%–41% in precision and 8%–34% in recall.