Author

Lina Sherkawi, Nada Ghneim, Oumayma Al Dakkak

Published in

International Journal on Computer and Communications Networks, Computational Intelligence and Data Analytics, Volume 1, No. 1, 2017

Abstract

This paper presents a rule-based approach to detect Arabic Speech Act types. Our approach is based on the fact that Arabic language is rich in morphology and that its syntax is inherently rule–based. The importance of Arabic Speech Act type recognition comes from its wide variety of application domains, such as: speech synthesis, speech and emotion recognition and conversational analysis. In this work, a rule-based Expert System has been developed in a bootstrapping manner, to classify an utterance speech act type to one of the sixteen types (interrogation, exclamation, negation, etc.). Our system was tested on a corpus of about 1500 sentences and achieved an accuracy rate of 98.92%. The expert system was used to automatically annotate a corpus (sentence-level annotation) by assigning a tag to each sentence according to its speech act type.

Keywords: Bootstrapping, Corpus Annotation, Grammatical Classification, Sentence Type Recognition, Speech Act.

Link to read full paper

https://www.praiseworthyprize.org/jsm/index.php?journal=ireci&page=article&op=view&path%5B%5D=21050