Development of an Arabic conversational intelligent tutoring system for education of children with autism spectrum disorder

Aljameel, Sumayh Saud (2018) Development of an Arabic conversational intelligent tutoring system for education of children with autism spectrum disorder. Doctoral thesis (PhD), Manchester Metropolitan University.

Preview

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (23MB) | Preview

Abstract

Children with Autism Spectrum Disorder (ASD) are affected in different degrees in terms of their level of intellectual ability. Some people with Asperger syndrome or high functioning autism are very intelligent academically but they still have difficulties in social and communication skills. In recent years, many of these pupils are taught within mainstream schools. However, the process of facilitating their learning and participation remains a complex and poorly understood area of education. Although many teachers in mainstream schools are firmly committed to the principles of inclusive education, they do not feel that they have the necessary training and support to provide adequately for pupils with ASD. One solution for this problem is to use a virtual tutor to supplement the education of pupils with ASD in mainstream schools. This thesis describes research to develop a Novel Arabic Conversational Intelligent Tutoring System (CITS), called LANA, for children with ASD, which delivers topics related to the science subject by engaging with the user in Arabic language. The Visual, Auditory, and Kinaesthetic (VAK) learning style model is used in LANA to adapt to the children’s learning style by personalising the tutoring session. Development of an Arabic Conversational Agent has many challenges. Part of the challenge in building such a system is the requirement to deal with the grammatical features and the morphological nature of the Arabic language. The proposed novel architecture for LANA uses both pattern matching (PM) and a new Arabic short text similarity (STS) measure to extract facts from user’s responses to match rules in scripted conversation in a particular domain (Science). In this research, two prototypes of an Arabic CITS were developed (LANA-I) and (LANA-II). LANA-I was developed and evaluated with 24 neurotypical children to evaluate the effectiveness and robustness of the system engine. LANA-II was developed to enhance LANA-I by addressing spelling mistakes and words variation with prefix and suffix. Also in LANA-II, TEACCH method was added to the user interface to adapt the tutorial environment to the autistic students learning, and the knowledge base was expanded by adding a new tutorial. An evaluation methodology and experiment were designed to evaluate the enhanced components of LANA-II architecture. The results illustrated a statistically significant impact on the effectiveness of LANA-II engine when compared to LANA-I. In addition, the results indicated a statistically significant improvement on the autistic students learning gain with adapting to their learning styles indicating that LANA-II can be adapted to autistic children’s learning styles and enhance their learning.