Document Retrieval through Concept Hierarchy Formulation

Abstract

The enormous growth of the Internet and the widespread use of computer systems in general created very large collections of electronic documents, and methods existing so far have proved unable to handle the massive amount of unstructured documents. In this article we discuss a variant of document retrieval, where traditional indexing is augmented by concept hierarchy (composed by observing concept roles in each member of the document collection) in order to improve accuracy. In addition, the document model is introduced as a way to recognize the limits inherent in document retrieval where no complete text understanding is feasible.