Authors:

Anna Ospina Bedoya

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

This project explores ways to enhance Purdue University's AI chat assistant by enriching its knowledge base with structured educational content from external, high-quality sources. Specifically, it focuses on improving the assistant's ability to support programming-related queries in digital forestry courses-such as Fundamentals of Remote Sensing- by leveraging real-world examples and instructional materials beyond traditional curricula. The current implementation centers on developing custom web crawlers for YouTube and GitHub. These crawlers extract relevant content such as code repositories, README files, video transcriptions, and tutorial metadata. The gathered materials are processed through a semantic embedding pipeline and stored in a vector database to enable retrieval-augmented generation (RAG) when responding to user queries. While still in progress, this approach is showing promise in improving the contextual relevance and specificity of the assistant's responses. The system architecture is built on a modular FastAPI backend and includes domain-specific parsers, an embedding service, and a semantic search interface using ChromaDB. A functional prototype of an AI tutor is under development, and tools for expanding the dataset are also being designed. The framework is intended to be scalable and extensible, supporting future integration of additional content sources and customizable user agents. Keywords: Web Crawling; Retrieval-Augmented Generation (RAG); Semantic Embeddings; Domain-Specific Content

Source:

Purdue University / 2025

Topics:

No topics listed

Co-authors:

Anna Ospina Bedoya