Book Recommendation System

Bridging Books and Readers Through Semantic Search

Premise

This project builds a book recommendation system that maps books and reader inquiries into a shared embedding space for semantic search implementation. Project Gutenberg data was loaded into Google’s pretrained BERT model to create meaningful document and query embeddings, then clustered with UMAP dimensionality reduction and HDBSCAN density-based clustering to surface semantically related reads.

Reasoning

LLM-derived embeddings were used in this system to capture thematic and conceptual similarity. It was developed as part of graduate coursework (CS 5660) exploring machine learning for recommendation architectures.

Github Link

Output Representations