Muhammad Usman Zafar

Academic

Career

Verbatim text

Rehnuma
An interactive chatbot for admission inquiries in Roman Urdu
Roman Urdu is a resource-poor language and hasn’t been enough researched and worked upon by 
researchers and developers. No intelligent chatbot for Roman Urdu is developed yet. Mostly the 
existing chatbots understand only a limited number of static queries and give only a limited number 
of static responses.  
So, we developed an interactive and intelligent chatbot that understands and gives dynamic 
responses to students’ queries in Roman Urdu. Our chatbot can handle different spelling variations 
of Roman Urdu words, detect offensive language, and handle out-of-scope questions. Chatbot also 
asks follow-up questions if the query is unclear or not enough information is provided. It then 
extracts the required information from the query, uses that information to calculate or predict the 
answer, and then replies to the user.  
It also remembers the context of the conversation. We used the RASA framework for the 
development of our chatbot and we got average accuracy of around 94% and an average F1 score 
of around 84% for all the user intents. 
It handles questions/queries falling in more than 50 different categories which include closing 
merit, programs offered, merit list date, financial aid, entry test schedule, entry test pattern, sports, 
scholarship, admission procedure, admission requirement, fee structure, semester system info, 
refund policy, admission open, admission deadline, campus life, hostel info, transportation, grading 
system, societies, facilities and many more. 
 
 
 
 
 
 
 
 
Technology Used: 
python, pyCharm, RASA framework  
colab, github, React 
Supervisor Name: 
Mr. Umair Arshad 
Group Members:   
Muhammad Usman Zafar (i16 - 0430) 
Okasha Khan (i17 - 0176) 
Zain-ul-Abideen (i16 - 0064)

AI enrichment