Project Abstract

This project was driven by the lack of an instant support system at UPenn, where students and staff rely on email for help, often waiting hours or days for replies. Recognising this inefficiency, the work set out to create a real-time, AI-powered chatbot that could improve access to university information and reduce reliance on manual support. The research uniquely positions itself by developing a context-specific chatbot tailored to UPenn, using deep learning, natural language processing, and finite state machines to manage dynamic, memory-aware conversations. The core research question asked whether such a chatbot could provide fast, accurate, and scalable assistance in a higher education environment. The chatbot was developed using a feedforward neural network trained on data collected via web scraping from UPenn’s website. Text pre-processing included tokenisation, stemming, and transforming input into numerical format using the Bag of Words technique. A Flask-based web frontend allowed users to interact with the chatbot in real time. The system achieved high performance, with 96% accuracy, 95% precision, 93% recall, and an average response time of 1.2ms. These results confirm the chatbot’s effectiveness in handling university-related queries with high reliability and consistency. Ultimately, this research contributes a functional and scalable AI solution to academic support, showing that chatbots can significantly reduce manual workloads and provide immediate, context-aware responses, paving the way for broader adoption in educational environments.

Keywords: AI Chatbot for Education, AI-Driven Student Support, UPenn Chatbot

 Conference Details

Session: B

Location: Sir Stanley Clarke Auditorium at 13:30 15:30

Markers: Lu Zhang, Solmaz Safari

Course: BSc Computer Science 3yr FT

Future Plans: I’m looking for work