Abstract:

Customer support plays an important role in building up long term relationship between companies and the customers. Companies  used  various  medium  to provide support to their customer like emails, messages and phone calls. This process is time consuming and you need to hire a dedicated Customer Support team who manage the customer queries, and this adds delay in the responses, decreases the customer support quality, and extra overhead on the company for the training of new individuals in the team. To automate this process, we are developing a deep learning based generative chatbot in Urdu language which will entertain users queries. In the past chatbots are designed using hand written technique and used Information Retrieval (IR) and pattern matching technique to find answer from database which results in non-human like responses. To add human like feel, we will be using Transformer based seq2seq model which will work on attention mechanism to make chatbot more human like.

Committee:

• Dr. Agha Ali Raza (Advisor),
• Dr. Asim Karim