Publications
- Characterization of Tumblr articles by information flow pathway
89th SIG-KBS, The Japanese Society for Artificial Intelligence
Author: Nozomi NORI, Kiyohiro YAMAGUCHI, Yasuhiro HASHIMOTO, Yu CHEN, Hirotada OHASHI
Keyword: Characterization, Information pathway, Tree edit distance, Ordinal Correlation, Collaborative filtering, Tumblr
Abstract: Information flow pathways seem to refect its intermediate people's preferences and its contents. So it is expected that characteristics of information which reflect people's preferences and contents are extracted by analyzing these pathways. In this article, we show that calculating item-item similarities by information pathway is more useful for characterization of Tumblr articles than calculating by simple co-occurrence relations.