Đang chuẩn bị liên kết để tải về tài liệu:
Báo cáo khoa học: "Learning Semantic Categories from Clickthrough Logs"

As the web grows larger, knowledge acquisition from the web has gained increasing attention. In this paper, we propose using web search clickthrough logs to learn semantic categories. Experimental results show that the proposed method greatly outperforms previous work using only web search query logs. sition in both precision and recall. We cast semantic category acquisition from search logs as the task of learning labeled instances from few labeled seeds. To our knowledge this is the first study that exploits search clickthrough logs for semantic category learning | Learning Semantic Categories from Clickthrough Logs Mamoru Komachi Nara Institute of Science and Technology NAIST 8916-5 Takayama Ikoma Nara 630-0192 Japan mamoru-k@is.naist.jp Shimpei Makimoto and Kei Uchiumi and Manabu Sassano Yahoo Japan Corporation Midtown Tower 9-7-1 Akasaka Minato-ku Tokyo 107-6211 Japan smakimot kuchiumi msassano @yahoo-corp.jp Abstract As the web grows larger knowledge acquisition from the web has gained increasing attention. In this paper we propose using web search clickthrough logs to learn semantic categories. Experimental results show that the proposed method greatly outperforms previous work using only web search query logs. 1 Introduction Compared to other text resources search queries more directly reflect search users interests Silverstein et al. 1998 . Web search logs are getting a lot more attention lately as a source of information for applications such as targeted advertisement and query suggestion. However it may not be appropriate to use queries themselves because query strings are often too heterogeneous or inspecific to characterize the interests of the user population. Although it is not clear that query logs are the best source of learning semantic categories all the previous studies using web search logs rely on web search query logs. Therefore we propose to use web search clickthrough logs to learn semantic categories. Joachims 2002 developed a method that utilizes clickthrough logs for training ranking of search engines. A search clickthrough is a link which search users click when they see the result of their search. The intentions of two distinct search queries are likely to be similar if not identical when they have the same clickthrough. Search clickthrough logs are thus potentially useful for learnin semantic categories. Clickthrough logs have the additional advantage that they are available in abundance and can be stored at very low cost.1 Our proposed method employs search click 1As for data availability MSN .