Using Search Strategies and a Description Logic Paradigm with Conditional Preferences for Literature Search

Abstract

In this paper, we propose a novel query formalism for literature search, which is based on description logics and variable-strength conditional preferences. Query languages of current search engines are very restricted in their expressive power. There are scientific search engines on the web, however, that have valuable metadata about research publications, authors, organisations, and scientific events. We show that description logics with conditional preferences allow for a more powerful query language, which can exploit the available metadata more effectively than the current approaches do. In particular, our approach allows for expressing nearly all important information search strategies proposed by Bates. We also describe a dialogue-oriented interface to the new query formalism and discuss possible extensions.