Turkish question answering systems and human-robot interaction

This thesis focuses on the development of a dialogue system with Turkish speaking capability aimed at enhancing the communication skills of children with special education needs. Specifically, it aims to provide an effective educational method equipped with technological facilities for children who face challenges in social and emotional communication. The research began with a rule-based pilot application, measuring children's real-time responses to enhance the system's efficacy. In the second phase, a new method was adopted to optimize answer selection for the Turkish dialogue system, achieving a success rate of 84.4%. In the third phase, a method was developed to accurately detect the emotional states of the children, achieving a success rate of 89.9%. Additionally, a humanoid robot design was introduced in this project, and a prototype of the robot was printed. Interactions with this robot increased the average speaking durations of the children by 15 minutes and reduced concentration losses by 68%. The results demonstrate that education supported by technological tools can significantly improve the social interaction and communication skills of children with special education needs. Moreover, it has been determined that interactions with the robot boost children's self-confidence, aiding them in expressing themselves more comfortably. This study offers valuable insights, a technological approach, and strategies to teachers, families, and special education specialists on how to make the educational process more effective for children with special education requirements.