Qdrant Launches Breakthrough Pure Vector-Based Hybrid Search, Raising the Bar for RAG and AI Applications

Applications of AI

BERLIN & NEW YORK–(BUSINESS WIRE)–Qdrant, the leading high-performance open source vector database company, today announced the release of BM42, a pure vector-based hybrid search approach that delivers more accurate and efficient search for modern Search Augmentation Generation (RAG) applications. The BM42 search algorithm significantly outperforms traditional text-based search for RAG and AI applications.

BM42 offers businesses another choice beyond traditional text search or traditional vector search. This pure vector-based hybrid search combines the best of both worlds to deliver better results at lower costs in the RAG space, giving users an edge in the AI-centric world.

Moving from keyword to vector-first search

Traditional keyword-based search engines using algorithms such as BM25, which are over 50 years old, are not optimized for the precise searches required for modern applications and have difficulty addressing certain RAG demands, especially for short text segments that require additional context for successful search and retrieval.

“By moving from keyword-based search to a fully vector-based approach, Qdrant has set a new industry standard,” said Andrey Vasnetsov, CTO and co-founder of Qdrant. “BM42 is more flexible, accurate and efficient for short pieces of text, which are more prominent in RAG scenarios, as it provides vector context in addition to the efficiency of traditional text search approaches. As Qdrant envisions a future centered around vector-based search, this release makes vector search more universally applicable, marking an important step towards the inevitable transition to a vector-first approach.”

– Message from our partners –

Qdrant's BM42 introduces a new way to classify search results, making it ideal for RAG applications. Unlike traditional keyword-based search, which is suitable for long-form content, Qdrant's solution integrates sparse and dense vectors to accurately identify relevant information within a document. Sparse vectors handle exact term matching. Dense vectors handle semantic relevance and deeper meaning.

Increased accuracy, efficiency and scalability

Developers are often faced with the critical decision of whether to choose sparse or dense vectors, or a hybrid approach. Many existing hybrid solutions suffer from scalability and accuracy issues, or are cost-prohibitive. Qdrant's new hybrid search system addresses these challenges, providing an efficient and cost-effective solution for both new and existing users. Most importantly, BM42 enables users to quickly move from prototype to production and scale their solutions globally.

For more details on the announcement, see: qdrant.tech/articles/bm42

About Qdrant

Qdrant is a high-performance, scalable open-source vector database and search engine that is essential for building the next generation of AI/ML applications. Qdrant can handle billions of vectors, supports semantically complex object matching, and is implemented in Rust for performance, memory safety, and scale. Recently, the company was named a top 10 startup in Sifted's 2024 B2B SaaS Rising 100, an annual ranking of the most promising European B2B SaaS startups valued under $1 billion.

contact address

For more information, please visit the Qdrant website or contact:

[email protected]

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *