Content-only querying structured contexts using formal concept analysis

Abstract

With the large increase of collections of structured documents (e.g. XML, HTML, ...), the need to retrieve different granules (fragments, sub-units, ...) of such documents, instead of the whole structure, becomes obvious. Nowadays, all existing Formal Concept Analysis-based Information Retrieval approaches address exclusively unstructured documents. They rely on the use of dyadic formal contexts (i.e. binary Documents × Terms relations). In this paper an original approach which consists of enlarging FCA-based IR paradigm to structured documents is proposed. Our approach stands from the idea of modeling structured documents by means of triadic formal contexts (i.e. ternary Documents × Terms × Structure relation). This allows to retrieve sub-units or fragments of structured documents. In structured information retrieval, queries may be of different types. This paper deals with content-only queries and gives a theoretical framework for both conjunctive as well as disjunctive forms.