---
title: "WeSearch: A Unified Model-Centric Retrieval Framework for WeChat Search"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.00530v1"
permalink: "https://feed.astralor.com/items/itm_e97803d69c39f74db0680ab1"
published: "2026-10-31T01:35:20.000Z"
score: 55
channel: "tech"
tags: ["推理效率","RAG","WeSearch","Unified Model-Centric Retrieval","Framework","WeChat Search","Prevailing","Semantic IDs"]
---

# WeSearch: A Unified Model-Centric Retrieval Framework for WeChat Search

## 速览

We present WeSearch, a unified, model-centric retrieval framework for large-scale, continuously evolving content search. Prevailing methods assign documents fixed hierarchical Semantic IDs, implicitly assuming each docu…

## 关键点

- We present WeSearch, a unified, model-centric retrieval framework for large-scale, continuously evolving content search.
- Prevailing methods assign documents fixed hierarchical Semantic IDs, implicitly assuming each document follows a single coherent semantic path-an assumption that often breaks down for multi-topic documents.
- WeSearch instead retrieves query-relevant clusters and then ranks atomic documents within the resulting query-specific candidate space, decoupling candidate pruning from fine-grained relevance modeling.

## 全文

We present WeSearch, a unified, model-centric retrieval framework for large-scale, continuously evolving content search. Prevailing methods assign documents fixed hierarchical Semantic IDs, implicitly assuming each document follows a single coherent semantic path-an assumption that often breaks down for multi-topic documents. WeSearch instead retrieves query-relevant clusters and then ranks atomic documents within the resulting query-specific candidate space, decoupling candidate pruning from fine-grained relevance modeling. Combined with relevance-aware training, compact document representations, and incremental index updates, this design enables weekly-or daily when needed-model refreshes. Deployed as the primary semantic retrieval path in WeChat Search alongside a lexical syntax tree, WeSearch shows consistent one-week A/B gains on corpora from hundreds of millions to billions of documents, and a one-month full-traffic A/B test confirms stable improvements of +0.68% CTR and +2.27% per-user article clicks, together with a 41% reduction in retrieval resource cost and a 19 ms reduction in graphic-text retrieval latency. WeSearch is now fully deployed in WeChat Search, serving 100% of production traffic.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_e97803d69c39f74db0680ab1