Query Expansion Based on Formal Concept Analysis From Retrieved Documents

In this paper, we propose a new formal concept analysis (FCA)-based query expansion approach, which uses the set of retrieved document collection against the whole document set. In this approach, description topics (DTs) are extracted from the documents and organized to denote precisely the user's information need. For a new query, we build a concept lattice from the extracted DTs, using the retrieved document collection as the formal context, and choose the most probable interpretations as query concepts. Our experiments are performed on two collections (data sets from TREC-7, TREC-8 and AP89). The experimental evaluation shows that our approach can reduce the overall computational overhead, and is as good as some typical query expansion approaches.