The dozens of existing search tools and the keyword-based search model have become the main issues of accessing the ever growing WWW. Various ranking algorithms, which are used to evaluate the relevance of documents to the query, have turn out to be impractical. This is because the information given by the user is too few to give good estimation. In this paper, we propose a new idea of searching under the multi-engine search architecture to overcome the problems. These include clustering of the search results and extraction of co-occurrence keywords which with the user's feedback better refines the query in the searching process. Besides, our system also provides the construction of the concept space to gradually customize the search tool to fit the usage for the user at the same time.
- Information extraction
- User feedback