Web knowledge extraction for improving search
Katsumi Tanaka · 2008
In this paper, we describe the extraction of Web knowledge from Web search engines and their usage for improving conventional Web search, such as supporting user's recall of vocabulary (query keywords), evaluation (quality, trustworthiness etc.) of search results, and re-ranking of search results. The extraction is based on the ideas of the "structural term co-occurrence" relationships and aggregating social annotation information, which can be easily computed by conventional Web search engines. Also, we describe a database interface to extract knowledge from Web search engines, which wraps accesses to Web search engines as virtual relational tables.