Ontology mediated information extraction with MASTRO SYSTEM-T

Abstract

In several data-centric application domains, the need arises to extract valuable information from unstructured text documents. The recent paradigm of Ontology Mediated Information Extraction (OMIE) faces this problem by taking into account the knowledge expressed by a domain ontology, and reasoning over it to improve the quality of extracted data. MASTRO SYSTEM-T is a novel tool for OMIE, developed by Sapienza University and IBM Almaden Research. In this work, we demonstrate its usage for information extraction over real-world financial text documents from the U.S. EDGAR system.