Intelligent Sinhala Question and Answer System by Incorporating Visual Clues

This research aims to develop an intelligent Sinhala question-and-answer system to enhance technological resources for Sinhala language processing. This AI-driven system will provide an interactive platform for college students to engage in conversational, object-based learning in Sinhala. Voice input will facilitate questions, and responses will be delivered in Sinhala audio, enhancing immersion and usability. This manages a local Mongo database for knowledge storage and retrieval of the data system that wants access to relevant information. This interactive Sinhala Q&A system used BERT model achitecute and it is trained on Sinhala dataset. Overall system has capability to adapt according to objects available in the vision field and then generate responses based on the object specific context. Object identification is done by YOLO model. In addition to developing a Sinhala BERT NLP model, Sinhala NLP Wordpiece tokenizer and Sinhala Dataset suitable for Sinhala BERT model are developed. Following result of this research can show that this system is an effective tool. When showing Banana to the system and asking ? (What are bananas?), system voice output will be (a fruit). In conclusion, this research will contribute to the field of educational technology and by developing this system, college students will have access to a tool that can assist them in improving their knowledge and proficiency in the Sinhala language.