Sease | Information Retrieval Applied

In our latest case study, we share how we helped the Bank for International Settlements – BIS improve data discovery by combining Information Retrieval techniques with AI-powered search, making complex statistical datasets easier to explore through natural language. If you're interested in search, AI, or data discovery, you might find it an interesting read: https://lnkd.in/eTYZ2vXm

If you decided that staying home under the air conditioning was a better idea than facing London's summer heat... we don't blame you. 😄 That's exactly why we've published the recording of the first talk from our latest London Information Retrieval & AI Meetup. 🎙️ Binary Quantization 101 by 🎙️Carly Richmond is now available on YouTube. Happy Watching: https://lnkd.in/emaWd2KK

Over the past few months, we’ve been able to dedicate focused engineering time to upstream a lot of work in Apache Solr. This work was sponsored by Bloomberg, and we're grateful for the support and collaboration that made it possible. In this post, we've collected a summary of the contributions we've completed recently, covering topics such as filtered vector search with ACORN, seeded KNN, early termination, quantization, Text-to-Vector improvements, and documentation updates. A huge thank you to everyone involved in this effort, and especially to Andrey Ukhanov, Ken LaPorte and Kevin Liang for their support and collaboration. Check the blogpost: https://lnkd.in/dbT2FBx7

Today's the day! 👏 The next London Information Retrieval & AI Meetup is happening this evening, both in London and online. We'll have two talks covering some of the most interesting topics in modern Information Retrieval: "Binary Quantization 101" - Carly Richmond, Principal Developer Advocate @ Elastic "From RAG to Agents: Building AI Applications on OpenSearch" - Itamar Syn-Hershko, CTO & Founder @ BigData Boutique 📍 London & Online 📍 Today, June 23rd See you later! https://lnkd.in/dnccxWs9

What AI and vector search features are available in #Vespa 8.626.55? In this blog post, we explore the AI side of the Vespa search engine, covering capabilities such as text and multimodal vectorisation, vector search capabilities, multi-vector search, learned sparse retrieval, cross-encoders and late interaction models reranking, hybrid search, Retrieval Augmented Generation, document enrichment through LLMs, ACORN-1, Adaptive Beam Search and many more. The article also discusses how these features can be configured and used within Vespa. Check it out: https://lnkd.in/dyw6wJmZ

At our next London Information Retrieval & AI Meetup on June 23rd, Itamar Syn-Hershko will host a session on how #OpenSearch can be used as a foundation for modern AI and agentic applications. The session will include real-world examples, reference architectures, and operational considerations for running OpenSearch at scale! 📍 London & Online 📅 June 23rd ➡️ Join us: https://lnkd.in/dWbtXnE5

A special milestone for our team 🎉 We are proud to announce that Anna Ruggero has become an Apache Solr Committer! This recognition reflects years of contributions to the project and active participation in the Solr community, helping advance one of the most widely used open source search technologies. Congratulations Anna! We can't wait to see what comes next 🚀

When using Lucene’s parent-child block-join mode, you are dealing with a “one-to-many relationship”. You have a single document (parent) owning multiple vector embeddings (children). To search these, Apache Lucene uses the DiversifyingChildrenFloatKnnVectorQuery. Its job is simple: find the k nearest parent documents by scoring children through HNSW graph traversal, returning at most one child — the best-scoring — per parent. 🔥 The reason why we started this in-depth code analysis and further implementation was to make the entire process faster. 🔥 Read more: https://lnkd.in/drRZJFSE

Curious about #BinaryQuantization? Join us on June 23rd for the next London Information Retrieval & AI Meetup, where Carly Richmond, Principal Developer Advocate at Elastic, will explore the world of binary quantization. You'll learn how it works, how it compares to other quantization approaches, and how Better Binary Quantization (BBQ) can be leveraged in #Elasticsearch! 📍 London & Online 📅 June 23rd ➡️ Register: https://lnkd.in/dnccxWs9

Apache Solr offers different ways to support faceting, but what impact do they have on performance? In this blog post, we explore the performance characteristics of #DocValues and #InvertedIndex when using the JSON Facet API in Apache Solr. The article walks through a series of benchmarks and discusses the results to help better understand the trade-offs between the two approaches. ➡️ Read more: https://lnkd.in/d_iH8_Z8