Ontology Driven Knowledge Extraction System with Application in e-Government

Abstract

Important sources of information are originally created in natural language. To make that knowledge computer processable it is necessary to understand the structure of natural languages, by adding lexical and syntactic information; to have a rich representation to encode the knowledge of sentences, like ontologies; and to develop algorithms to bridge the gap between natural languages and computer processable representations. In this paper we present the architecture, modules and results of a prototype that uses an ontology to represent the world concepts and their relationships, and also to guide the process of extracting information from natural language documents. The system was tested using minutes of Portuguese municipalities’ meetings. Initial results are presented for three topics of municipalities' affairs: the subsidies granted, the building permits requested, and the existing protocols with other institutions.