Home // ALLDATA 2016, The Second International Conference on Big Data, Small Data, Linked Data and Open Data // View article

An Ontology-based Method for Discovering Specific Concepts from Texts via Knowledge Completion

Authors:
Céline Alec
Chantal Reynaud-Delaître
Brigitte Safar

Keywords: discovering concepts from texts; ontology enrichment; ontology population.

Abstract:
Heterogeneity in user's queries and data sources can easily cause problems in perceiving sufficient information to form correct answers. In this paper, we address this issue when data sources are unstructured short texts describing only key characteristics of concerned individuals but when keywords in user's queries are customized concepts. To bridge the gap between texts and user's concepts, we propose an ontology-based approach, named SAUPODOC (Semantic Annotation Using Population of Ontology and Definition of Classes), to discover formal definitions of specific concepts via population of property assertions. Property assertions are extracted from texts but the texts under our consideration are incomplete, i.e., information about the target concepts is missing. To solve this problem, we further propose a method to extract property assertions by exploiting LOD (Linked Open Data) datasets to deal with missing and multiple values. Experiments have been carried out in two application domains, whose results show a clear benefit of SAUPODOC over well-known classifiers.

Pages: 96 to 101

Copyright: Copyright (c) IARIA, 2016

Publication date: February 21, 2016

Published in: conference

ISSN: 2519-8386

ISBN: 978-1-61208-457-2

Location: Lisbon, Portugal

Dates: from February 21, 2016 to February 25, 2016