A term-based methodology for template creation in information extraction

Abstract

In this paper, we are concerned with the problem of auto-matic template creation for Information Extraction (IE) and we present a methodology for the creation of IE templates. Our approach proposes the semi-Automatic construction of a semantic representation of textual information based on recognition of multi-word and nested terms and Named Entities (NEs) and subsequent exploitation of term and NE con-text for the induction of Information Extraction template rules.