FCA and IR: The Story So Far

                          Claudio Carpineto

                    Fondazione Ugo Bordoni, Rome
                          carpinet@fub.it



Abstract. The application of Formal Concept Analysis (FCA) to Infor-
mation Retrieval (IR) is twenty-five years old. Over this period, a number
of papers have explored the potentials of FCA for various information
finding tasks while several system prototypes have been made available
for experimentation and testing. In this talk we survey what has been
achieved so far, discussing lessons and implications from a few successful
case studies, including browsing of web search results, smooth integration
of data driven and model driven search, exploratory document mining,
and semantic text classification. We argue that, in spite of these good
results, the impact of FCA on mainstream IR has been limited, due to
theoretical and practical limitations. Nowadays, the tremendous increase
in the richness and diversity of electronic data along with the inherent
shortcomings of traditional search engines call for better IR techniques,
opening up new opportunities for knowledge-intensive methods. Integra-
tion of FCA with existing search technology and new theoretical advances
capable of extending the applicability of FCA to IR beyond the Boolean
retrieval model are seen as key factors for the further development of
this field.