Home // SEMAPRO 2012, The Sixth International Conference on Advances in Semantic Processing // View article

Comparing a Rule-Based and a Machine Learning Approach for Semantic Analysis

Authors:
François-Xavier Desmarais
Michel Gagnon
Amal Zouaq

Keywords: Semantic role labeling; evaluation; rule-base systems; machine learning.

Abstract:
Semantic analysis is a very important part of natural language processing that often relies on statistical models and machine learning approaches. However, these approaches require resources that are costly to acquire. This paper describes our experiments to compare Anasem, a Prolog rule-based semantic analyzer, with the best system of the Conference on Natural Language Learning (CoNLL) shared task dedicated to a sub-task of semantic analysis: Semantic Role Labeling. Both CoNLL best system and Anasem are based on a dependency grammar, but the major difference is how the two systems extract their semantic structures (rules versus machine learning). Our results show that a rule-based approach might still be a promising solution able to compete with a machine learning system under certain conditions.

Pages: 103 to 108

Copyright: Copyright (c) IARIA, 2012

Publication date: September 23, 2012

Published in: conference

ISSN: 2308-4510

ISBN: 978-1-61208-240-0

Location: Barcelona, Spain

Dates: from September 23, 2012 to September 28, 2012