Document knowledge representation using description logics for information extraction and querying

The document representation is an important aspect of both information retrieval and information extraction. In this paper the bag of words representation of documents is enriched with lexical, conceptual and contextual relationships. In order to effectively represent and inference with this representation, a mathematical model is required. In this work, the enhanced interrelated set of bag of words that is semantically lattice of words is represented using description logic. The conceptual taxonomy extracted using WordNet can be represented naturally using DL. The DL also provides consistency, satisfiability, instance checking, subsumption services to semantically extend the initial enhanced lattice of words. The contextual knowledge can be extracted by syntactic patterns, dependency relations and heuristics. Tourist and sports domains are taken for the implementation of this work. This knowledge extracted and represented can be used in applications like information retrieval and information extraction.