Context Comparison of Bursty Events in Web Search and Online Media

Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing, pages 1077–1087, MIT, Massachusetts, USA, 9-11 October 2010. c⃝2010 Association for Computational Linguistics Context Comparison of Bursty Events in Web Search and Online Media Yunliang Jiang University of Illinois Urbana, IL, 61801 jiang8@illinois.edu Cindy Xide Lin University of Illinois Urbana, IL, 61801 xidelin2@illinois.edu Qiaozhu Mei University of Michigan Ann Arbor, MI, 48109 qmei@umich.edu Abstract In this paper, we conducted a systematic com- parative analysis of language in different con- texts of bursty topics, including web search, news media, blogging, and social bookmark- ing. We analyze (1) the content similarity and predictability between contexts, (2) the cov- erage of search content by each context, and (3) the intrinsic coherence of information in each context. Our experiments show that so- cial bookmarking is a better predictor to the bursty search queries, but news media and so- cial blogging media have a much more com- pelling coverage. This comparison provides insights on how the search behaviors and so- cial information sharing behaviors of users are correlated to the professional news media in the context of bursty events. 1 Introduction Search is easy. Every day people are repeating the queries they have used before, trying to access the same web pages. A smart search engine tracks the preference and returns it next time when it sees the same query. When I search for “msr” I always try to access Microsoft research; and even if I misspelled it, a smart search engine could suggest a correct query based on my query history, the current ses- sion of queries, and/or the queries that other people have been using. Search is hard. I search for “social computing” because there was such a new program in NSF; but the search engine might have not yet noticed that. People use “msg” to access monosodium glutamate in most of the cases, but tonight there is a big game in Madison square garden. H1N1 suddenly became a hot topic, followed by a burst of the rumor that it was a hoax, and then the vaccine. The informa- tion need of users changed dramatically during such a period. When a new event happens, the burst of new contents and new interests make it hard to pre- dict what people would search and to suggest what queries they should use. Web search is easy when the information need of the users is stable and when we have enough histor- ical clicks. It becomes much more difficult when a new information need knocks the door or when there is a sudden change of the information need. Such a shift of the information need is usually caused by a burst of new events or new interests. When we are lack of enough historical observa- tions, why don’t we seek help from other sources? A bursting event will not only influence what we search, but hopefully also affect what we read, what we write, and what we tag. Indeed, there is al- ready considerable …

Content truncated to 500 words