FURI | Spring 2025

ESHA: Enhanced SHopping Assistance

Data icon, disabled. Four grey bars arranged like a vertical bar chart.

This research enhances online shopping by improving how search engines understand user queries. Traditional search methods often miss the context and intent behind natural language input. This project applies diffusion models and large language models (LLMs) to convert both product data and user queries into a shared vector space, allowing for more accurate results. Evaluation through ranking metrics such as hit rate and normalized discounted cumulative gain (nDCG) demonstrates improved search performance. The outcome supports a more intuitive, efficient, and satisfying e-commerce experience.