Conversational AI model to answer Covid-related queries
Final Year Project student Xin Hui Wang developed a model to respond to user inputs
Under the supervision of Dr Negender Aneja, Final Year Project student Xin Hui Wang developed a chat system that applied attention mechanism in recurrent neural network to respond to COVID-19 inquiries. Three types of Luong’s scoring methods in attention mechanisms were investigated; the Dot Attention Mechanism, the General Attention Mechanism and the Concat Attention Mechanism. Based on the experimental results, the dot attention mechanism achieved the highest accuracy of 87% when evaluated on test questions obtained directly from the database. When asked questions with natural variations, a 63% human verification accuracy was obtained, compared to 16% accuracy when no attention mechanism was used. Based on these results, the paper demonstrate that chatbots can potentially be used everywhere with round-the-clock accessibility. This work has been published in the International Journal of Intelligent Networks.
—
Universiti Brunei Darussalam
Jalan Tungku Link, BE1410
Negara Brunei Darussalam
office.sds@ubd.edu.bn