Opinions in Federated Search: University of Lugano at TREC 2014 Federated Web Search Track.

摘要

This technical report presents the work carried out at the University of Lugano on TREC 2014 Federated Web Search track. The main motivation behind our approach is to provide better coverage of opinions that are present in federated resources. On the resource selection and vertical selection steps, we apply opinion mining to select opinionated resources/ verticals given a user's query. We do this by combining relevance-based selection with lexicon-based opinion mining. On the results merging step, we diversify the final document ranking based on sentiment using the retrieval-interpolated diversification method.