From c581b96f786ccab9d103df1a91e5a070bcd32d5a Mon Sep 17 00:00:00 2001 From: "jiliang.ljl" Date: Mon, 13 Jul 2026 20:45:16 +0800 Subject: [PATCH] docs: clarify multi-query wording --- content/blog/en/2026-06-12-zvec-release.mdx | 4 ++-- content/blog/en/2026-07-07-zvec-fts.mdx | 2 +- content/blog/zh/2026-06-12-zvec-release.mdx | 2 +- content/blog/zh/2026-07-07-zvec-fts.mdx | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/content/blog/en/2026-06-12-zvec-release.mdx b/content/blog/en/2026-06-12-zvec-release.mdx index e0dbf52ce..5e4a6b71e 100644 --- a/content/blog/en/2026-06-12-zvec-release.mdx +++ b/content/blog/en/2026-06-12-zvec-release.mdx @@ -13,7 +13,7 @@ You can find the complete [release notes on GitHub](https://github.com/alibaba/z ## Native Full-Text Search & Hybrid Retrieval -Zvec v0.5.0 introduces native [**full-text search**](https://zvec.org/en/docs/db/concepts/fts-index/). You can now attach an FTS index to any string field via `FtsIndexParam`, and query it with natural-language match strings or structured query expressions — no external search engine required. With FTS + vector hybrid retrieval in `MultiQuery`, text and vector signals can be combined in a single query, making Zvec a complete in-process hybrid-retrieval engine. +Zvec v0.5.0 introduces native [**full-text search**](https://zvec.org/en/docs/db/concepts/fts-index/). You can now attach an FTS index to any string field via `FtsIndexParam`, and query it with natural-language match strings or structured query expressions — no external search engine required. With FTS + vector hybrid retrieval in a single query call, text and vector signals can be combined and reranked together, making Zvec a complete in-process hybrid-retrieval engine. **Define an FTS-indexed field:** @@ -152,7 +152,7 @@ print(result) --- -## Unified Query API & Native Multi Query +## Unified Query API & Native Multi-Query Execution The query surface is now simpler and faster. `VectorQuery` has been unified into a single `Query` type — one consistent entry point for vector search, ID lookup, and FTS. `VectorQuery` remains as a deprecated alias and will be removed in a future release, so please migrate to `Query`. diff --git a/content/blog/en/2026-07-07-zvec-fts.mdx b/content/blog/en/2026-07-07-zvec-fts.mdx index b3ca4542e..7675f0763 100644 --- a/content/blog/en/2026-07-07-zvec-fts.mdx +++ b/content/blog/en/2026-07-07-zvec-fts.mdx @@ -289,7 +289,7 @@ For users, it provides: + `match_string` for natural-language keyword queries; + `query_string` for Boolean and phrase queries; + FTS-only Collection for pure-text search; -+ MultiQuery + RRF / Weighted reranker for FTS x Vector hybrid search. ++ FTS–vector hybrid search with multi-path retrieval and RRF or weighted reranking. Under the hood, it reuses Zvec's Segment lifecycle, persistence, recovery, deletion filtering, and compaction mechanisms, and integrates keyword search into the existing database execution framework via the Roaring write format, BitPacked read-only format, BM25, WAND, and two-phase phrase verification. diff --git a/content/blog/zh/2026-06-12-zvec-release.mdx b/content/blog/zh/2026-06-12-zvec-release.mdx index d2809ee50..186ccb959 100644 --- a/content/blog/zh/2026-06-12-zvec-release.mdx +++ b/content/blog/zh/2026-06-12-zvec-release.mdx @@ -14,7 +14,7 @@ description: 🎉 版本发布 ## 原生全文检索与混合检索 -Zvec v0.5.0 引入了原生[**全文检索**](https://zvec.org/zh/docs/db/concepts/fts-index/)。现在你可以通过 `FtsIndexParam` 为任意字符串字段挂载 FTS 索引，并使用自然语言匹配串或结构化查询表达式进行检索——无需依赖外部搜索引擎。借助 `MultiQuery` 中的 FTS + 向量混合检索，文本与向量信号可在单次查询中组合，使 Zvec 成为完整的进程内混合检索引擎。 +Zvec v0.5.0 引入了原生[**全文检索**](https://zvec.org/zh/docs/db/concepts/fts-index/)。现在你可以通过 `FtsIndexParam` 为任意字符串字段挂载 FTS 索引，并使用自然语言匹配串或结构化查询表达式进行检索——无需依赖外部搜索引擎。通过在单次查询调用中融合 FTS 与向量检索，文本与向量信号可以统一召回和重排，使 Zvec 成为完整的进程内混合检索引擎。 **定义带 FTS 索引的字段：** diff --git a/content/blog/zh/2026-07-07-zvec-fts.mdx b/content/blog/zh/2026-07-07-zvec-fts.mdx index 2c2a94583..0cb44395e 100644 --- a/content/blog/zh/2026-07-07-zvec-fts.mdx +++ b/content/blog/zh/2026-07-07-zvec-fts.mdx @@ -289,7 +289,7 @@ Zvec 的 FTS 可以概括为一句话：**把全文检索做成向量数据库 + `match_string` 覆盖自然语言关键词查询； + `query_string` 支持布尔和短语查询； + FTS-only Collection 支持纯文本搜索； -+ MultiQuery + RRF / Weighted reranker 支持 FTS x Vector 混合检索。 ++ 通过多路召回与 RRF 或加权重排实现 FTS–向量混合检索。 对系统来说，它复用了 Zvec 的 Segment 生命周期、持久化、恢复、删除过滤和 compaction 机制；在内部通过 Roaring 写入格式、BitPacked 只读格式、BM25、WAND 和短语两阶段验证，把关键词检索融入原有数据库执行框架。