Eduowl

Python, OpenAI API, LangChain, Web Scraping, NLP

Eduowl was created to simplify the university admissions journey by guiding students through the application process and helping them discover academic majors that align with their interests, strengths, and career goals.

Github repository

Inspiration

As a group of college students who vividly recall the complexities and uncertainties of the university admissions process, we found our collective inspiration. Our team of four, each from different majors and backgrounds, recognized a common challenge faced by high school graduates: the overwhelming task of choosing a major while navigating complex admissions requirements. Eduowl was created to turn this daunting pre-college experience into an empowering and streamlined journey, particularly for students applying to Rowan University.

What it does

Eduowl revolutionizes the university admissions process by offering two core capabilities:

• Admissions Guidance Chatbot: An AI-driven chat interface that answers questions about Rowan University, including academics, admissions requirements, campus life, and student resources.
• Major Recommendation System: Evaluates students’ interests, strengths, and career goals to recommend suitable academic majors.
• Interactive AI Conversations: Provides engaging, real-time, context-aware responses.
• Data-Driven Advising: Uses structured admissions data sourced directly from official university resources.

How we built it

• Backend: Python
• Web Scraping: BeautifulSoup, Requests
• PDF Processing: PyPDF2
• AI & NLP: OpenAI API
• Framework: LangChain
• Conversation Memory: ConversationBufferMemory
• Data Handling: Text chunking and token estimation

Building Eduowl began by gathering the latest admissions information from Rowan University’s official website. We implemented web scraping techniques to extract accurate and up-to-date admissions data, with a focus on international admissions requirements. To extend the chatbot’s knowledge base, we processed PDF documents using PyPDF2.

The chatbot was powered by OpenAI’s language models and integrated using LangChain. ConversationBufferMemory enabled the bot to maintain context across interactions, resulting in coherent and relevant responses. Text chunking and token estimation were used to efficiently handle large volumes of scraped and document-based content.

Demo

Challenges we ran into

• Persistent errors when attempting to integrate JSON data into Azure AI Studio
• Handling large volumes of scraped and PDF-based data
• Maintaining response accuracy and contextual relevance

Accomplishments

• Built a functional AI-powered admissions chatbot
• Designed a form-based major recommendation feature
• Successfully pivoted from Azure AI Studio to the OpenAI API
• Delivered a student-focused solution addressing real admissions challenges

What we learned

• How to build conversational AI using OpenAI and LangChain
• How to scrape, structure, and process real-world admissions data
• The importance of adaptability when encountering technical roadblocks

What is next for Eduowl

• Expand major recommendations across additional academic disciplines
• Integrate Microsoft Azure to enhance AI capabilities
• Improve chatbot memory, intelligence, and response accuracy
• Add academic advising and writing assistance features

• 2025 3
• 2024 4
• 2023 2
• 2022 1