Current search technologies work in "one size fits all" fashion. Therefore, the answer to a query is independent of specific user information need. In this paper, we des...
In this paper, we propose a new methodology based on directed graphs and the TextRank algorithm to automatically induce general-specific noun relations from web corpora frequency c...
This paper illustrates a ranking scheme which combines fulltext, anchor text and URL structure for homepage finding in hybrid peer-to-peer networks. The experimental results show...
Abstract Most ranking algorithms are based on the optimization of some loss functions, such as the pairwise loss. However, these loss functions are often different from the criter...
Given a database with missing or uncertain content, our goal is to correct and fill the database by extracting specific information from a large corpus such as the Web, and to d...