We show that incorporating user behavior data can significantly improve ordering of top results in real web search setting. We examine alternatives for incorporating feedback into...
We consider the problem of combining ranking results from various sources. In the context of the Web, the main applications include building meta-search engines, combining ranking...
Cynthia Dwork, Ravi Kumar, Moni Naor, D. Sivakumar
The context of a search query often provides a search engine meaningful hints for answering the current query better. Previous studies on context-aware search were either focused ...
SimRank has been considered as one of the promising link-based ranking algorithms to evaluate similarities of web documents in many modern search engines. In this paper, we investi...
Beam search is used to maintain tractability in large search spaces at the expense of completeness and optimality. We study supervised learning of linear ranking functions for con...