https://qdrant.tech/documentation/search-patterns/Recent content in Search Patterns on Qdrant - Vector Search EngineHugoen-usinfo@qdrant.tech (Andrey Vasnetsov)info@qdrant.tech (Andrey Vasnetsov)https://qdrant.tech/documentation/search-patterns/choose-embedding-model/Tue, 15 Jul 2025 00:00:00 +0000info@qdrant.tech (Andrey Vasnetsov)https://qdrant.tech/documentation/search-patterns/choose-embedding-model/<h1 id="how-to-choose-an-embedding-model-evaluation--tradeoffs">How to Choose an Embedding Model: Evaluation &amp; Tradeoffs</h1> <p>No matter if you are just beginning your journey in the world of vector search, or you are a seasoned practitioner, you have probably wondered how to choose the right embedding model to achieve the best search quality. There are some public benchmarks, such as <a href="https://huggingface.co/spaces/mteb/leaderboard" target="_blank" rel="noopener nofollow">MTEB</a>, that can help you narrow down the options, but datasets used in those benchmarks will rarely be representative of your domain-specific data. Moreover, search quality is not the only requirement you could have. For example, some of the best models might be amazingly accurate for retrieval, but you can&rsquo;t afford to run them, e.g., due to high resource usage or your budget constraints.</p>https://qdrant.tech/documentation/search-patterns/query-decomposition/Mon, 01 Jan 0001 00:00:00 +0000info@qdrant.tech (Andrey Vasnetsov)https://qdrant.tech/documentation/search-patterns/query-decomposition/<h1 id="query-decomposition-for-multi-hop-questions">Query Decomposition for Multi-Hop Questions</h1> <table> <thead> <tr> <th>Time: 15 min</th> <th>Level: Intermediate</th> <th></th> <th></th> </tr> </thead> <tbody> </tbody> </table> <p>A multi-hop question chains two facts: the second depends on the answer to the first. &ldquo;Where was the director of the film Inception born?&rdquo; needs the director, then that person&rsquo;s birthplace. A single query retrieves chunks about the film, but the birthplace sits in a chunk about Christopher Nolan that never mentions Inception. Reranking and fusion only reorder what one query already retrieved, so they can&rsquo;t recover evidence that was never in the candidate set.</p>https://qdrant.tech/documentation/search-patterns/vector-search-filtering/Tue, 10 Sep 2024 00:00:00 +0000info@qdrant.tech (Andrey Vasnetsov)https://qdrant.tech/documentation/search-patterns/vector-search-filtering/<h1 id="a-complete-guide-to-filtering-in-vector-search">A Complete Guide to Filtering in Vector Search</h1> <p>Imagine you sell computer hardware. To help shoppers easily find products on your website, you need to have a <strong>user-friendly <a href="https://qdrant.tech" target="_blank" rel="noopener nofollow">search engine</a></strong>.</p> <p><img src="https://qdrant.tech/articles_data/vector-search-filtering/vector-search-ecommerce.png" alt="vector-search-ecommerce"></p> <p>If you’re selling computers and have extensive data on laptops, desktops, and accessories, your search feature should guide customers to the exact device they want - or at least a <strong>very similar</strong> match.</p> <p>When storing data in Qdrant, each product is a point, consisting of an <code>id</code>, a <code>vector</code> and <code>payload</code>:</p>