10x the Retrieval Quality of Vector Search - Enhancing Semantic Search and RAG Systems
Data Science Festival via YouTube
Overview
Discover how to significantly enhance vector search retrieval quality in this 43-minute talk by György Móra from Superlinked, presented at the Data Science Festival. Learn techniques to improve semantic search, RAG, and recommender systems by incorporating more information into vectors, potentially increasing quality up to tenfold. Explore the challenges of encoding and combining different data types into a single vector, and see how Superlinked's platform enables interactive experimentation and seamless production deployment. Gain insights on simplifying the transition from notebook experiments to production, avoiding post-processing and reranking through multi-modal vectors, and implementing five key building blocks to validate prototypes without complex pipelines. Suitable for technical practitioners, this session from the Data Science Festival MayDay event 2024 offers valuable knowledge for optimizing vector-based retrieval systems.
Syllabus
10x the Retrieval Quality of your Vector Search - Data Science Festival
Taught by
Data Science Festival