8× faster binary search：从编译代码一路挖到机械同理心 这篇性能文章很适合给写 Rust 的人当一堂“CPU 现实主义”复习课。作者面对的不是玩具问题，而是 scikit-learn 直方图梯度提升里一个很实际的步骤：把大批浮点数分桶到 0–254...

本文讨论了Milvus 2.6.x版本向量检索引擎中通过多级归并树实现高效搜索的方法，重点介绍了查询路径、Query Delegator的作用及一致性模型对搜索延迟的影响，并强调了分布式搜索的复杂性和一致性管理的重要性。

本文讨论了Milvus 2.6.x的数据处理架构，重点介绍了Growing和Sealed段的查询机制。Growing段由Streaming Node处理，实时更新；Sealed段由Query Node加载历史数据。Query Delegator负责生成查询计划并合并结果。Segcore作为段级执行层，与Knowhere索引内核协同工作，实现高效的向量检索。文章还探讨了查询过程中的数据加载、索引管理及动态更新与不可变索引之间的平衡。

SaaS software vendor HubSpot has described how its semantic search platform grew from a proof of concept into an internal service that now manages more than 20 billion vectors across 38-plus...

PostgreSQL forbids indexing with the default text search configuration because it's mutable—change the setting, and your index silently disagrees with reality.

Swiggy detailed real-time machine-learning ranking system for autocomplete built on OpenSearch. The architecture separates candidate generation and ranking, uses feature stores for real time...

pgvector is excellent. It is also, at large scale, expensive — because the HNSW index it gives you wants to live in memory to be fast, and “wants to live in memory” stops being a casual statement...

本文介绍了Claude的WebSearch和WebFetch工具。WebSearch支持网络搜索并返回格式化结果，WebFetch用于从指定URL抓取内容。两者均具备域名过滤和搜索次数限制功能，以确保信息的准确性和安全性。

本文介绍了Claude的WebSearch和WebFetch工具。WebSearch支持网络搜索、域名过滤和搜索次数限制，帮助用户获取最新信息。WebFetch用于从指定URL抓取内容，转换为markdown格式，并提取信息。两者均注重安全性和资源限制，确保高效安全的数据访问。

Synchronizing the Senses: Powering Multimodal Intelligence for Video SearchBy Meenakshi Jindal and Munya MarazanyeToday’s filmmakers capture more footage than ever to maximize their creative...

DoorDash has launched a multimodal machine learning system that aligns product images, text, and user queries in a shared embedding space. Trained on 32 million labeled query-product pairs using...

Circle to Search 更新后，用户可同时识别图像中的多个对象，提升搜索体验。用户只需圈选即可获取详细信息和相关链接，方便购物和获取灵感。该功能在三星 Galaxy S26 和 Pixel 10 上可用。

The AI Evolution of Graph Search at Netflix: From Structured Queries to Natural LanguageBy Alex Hutter and Bartosz BalukiewiczOur previous blog posts (part 1, part 2, part 3) detailed how...