KeanGlobal : An AI Assistant for University Information and Navigation​

David Vasquez

Co-Presenters: Ummu Yuzugulluer, Chin-Chien Lin, Noor Ul Ain Balouch

College: Hennings College of Science Mathematics and Technology

Major: MS.COMPUTER/SCIENCE

Faculty Research Mentor: Kwak, Daehan  

Abstract:

Universities provide substantial digital resources through websites, policy documents, and online service portals, but students frequently have trouble finding accurate, timely, and appropriate information about academic services, administrative procedures, and campus locations. This project introduces KeanGlobal, a multilingual, AI-powered campus assistance system aiming to assist Kean University students using a chat-first, human-centered interface. The technology allows students to ask questions using natural language about university policies, services, and campus locations while receiving conversational responses based on official institutional information. KeanGlobal is built as a full-stack application with a React UI and a FastAPI backend, which interacts with a locally hosted open-source language model using Mistral and Ollama. By implementing the language model locally, the system prioritizes privacy, data protection, and institutional control, setting it apart from cloud-based or third-party AI solutions. The architecture is intended to provide retrieval-augmented generation, which allows the AI to react with curated Kean University documents rather than generic online knowledge. In its current version, the system facilitates policy clarification and service requests, with plans to expand functionality to include context-aware campus navigation through interactive mapping technology. This project demonstrates how conversational AI-driven technologies can make higher education support systems easier to use and more accessible. KeanGlobal advances the field of educational technology and human-computer interaction research by demonstrating a useful, institutionally regulated method of AI-powered student support. Ongoing development focuses on expanding language support, boosting document coverage, incorporating vector-based retrieval, and offering administrative tools to aid system evaluation and improvement.Keywords: Conversational AI, Campus Navigation, Educational Technology, Human-Computer Interaction, University Support Systems