Product Discovery

Product discovery research aims to improve how users find desired items, whether makeup or scientific experimental designs, by enhancing search and recommendation systems. Current approaches leverage machine learning, particularly employing techniques like reinforcement learning to analyze agent behavior and extract meaningful "gadgets" (reusable subroutines) or using CLIP-like models for zero-shot comparative recommendations (e.g., finding similar items with different attributes). This work is significant for improving online shopping experiences and accelerating scientific discovery by providing more efficient and insightful tools for navigating vast datasets and complex problem spaces.