Web Intelligence and Intelligent Agent Technology, IEEE/WIC/ACM International Conference on (2011)
Aug. 22, 2011 to Aug. 27, 2011
We pose a new problem of discovering associations between events in our daily lives and their characteristic items, such as (Halloween, pumpkin) and (Christmas, chimney). To solve the problem, we dopted an approach similar to that of existing research on event detection, which tries to discover events by detecting bursts of occurrence frequency of a relevant term in a document stream, where the term (item) is associated with the discovered event. We extracted events from blog entries available on the Web, while the previous studies mostly used news articles as document streams. Blog entries are shown to have quite different characteristics to news articles. Considering this fact, we developed a method for discovering the associations by integrating existing techniques that can handle and take advantage of the characteristics of blog data. We verified through experiments using actual data that the proposed approach works quite well.
event extraction, burst detection, community detection
Shin-ya Sato, Masami Takahashi, Tetsuya Nakamura, Masato Matsuo, "Revealing Associations between Events and Their Characteristic Items", Web Intelligence and Intelligent Agent Technology, IEEE/WIC/ACM International Conference on, vol. 01, no. , pp. 243-248, 2011, doi:10.1109/WI-IAT.2011.135