Towards a Portable SLU System Applied to MSA and Low-resourced Algerian Dialects

Abstract

As the most used approach to extend a Spoken language Understanding (SLU) from a language to another, Machine translation achieves high performance for English domains, which is not the case for other languages, especially low-resourced ones as Arabic and its dialects. To avoid Machine Translation approach which requires huge parallel corpora, we will investigate, in this paper, the problem of user’s intent interpretation from natural language queries to a system’s semantic representation format across the languages and dialects, namely: English, Modern Standard Arabic (MSA) and four vernacular Algerian dialects from different regions: Blida, Djelfa, Tenes and Tizi-Ouzou. We should note that the domain we have chosen to run our experiments is a special application of school management. For this, We use three classifiers: kNN, Gaussian Naive Bayes and Bernoulli Naive Bayes which led to an average accuracy of 90%.