A joint link and content approach to information retrieval and distillation

Abstract

This paper presents a combined link and content based search method. A content relevance index is proposed to measure information quality. A sliding window model (SWM) is proposed to construct a dynamic graph and control the size of the node set of a graph. An exponential smoothing (ES) approach is applied to guide the search. Experimental results show that the proposed approach with the above features outperforms the link based approaches in terms of both information quality and computing time.