---
title: "AskChem：面向化学文献综合的声明为中心基础设施"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2607.28618v1"
permalink: "https://feed.astralor.com/items/itm_f2552ca7b9a24c7fc6fcac05"
published: "2026-07-30T17:59:11.000Z"
score: 78
channel: "ai"
tags: ["AI Agent","AskChem","化学文献检索","声明级检索","证据图谱"]
---

# AskChem：面向化学文献综合的声明为中心基础设施

## AI 点评

> 这篇论文最值得关注的地方是把「引用」从论文级细化到「声明级」，而且直接给 AI 代理开了 MCP 接口，等于在化学文献和 Agent 之间铺了一条可审计的证据管道。对做 RAG 或科学 QA 的同学来说，AskChem-Bench 的 DOI 可解析率指标值得直接借鉴，尤其是 100% 对比 88.3% 这个数字，说明检索粒度对答案可信度影响巨大。如果你在化学、材料或生命科学领域做知识库，这套思路和架构可以很快移植过来。

## 速览

化学文献综合常需从大量论文中拼凑具体发现，但现有检索系统只返回排序文档，科学家和 AI 代理需手动定位、验证并组装跨论文答案。AskChem 将检索单元从论文转为携带来源的声明，每篇论文被转化为原子化、类型化的声明，并由源 DOI、逐字引文或证据定位器支撑。AskChem 提供分面分类、证据图谱和动态主题分类等互补结构，当前索引来自 147K 篇论文的 240 万条声明，并通过 Web、REST、SDK 和 MCP 开放给 AI 代理。在 AskChem-Bench 测试中，基于 AskChem 的检索使 GPT-5.5 读者实现了 100% 可解析 DOI，而基线为 88.3%。

## 关键点

- AskChem 将检索单元从论文转换为带来源的原子化声明，每条声明由 DOI 和逐字引文/证据定位器支撑。
- 系统提供分层分面分类、证据图谱和动态生活分类三种互补结构，支持层级探索与关系查询。
- 当前索引 147K 篇论文中的 240 万条声明，提供 Web 界面及 REST、SDK、MCP 接口供 AI 代理使用。
- 在 AskChem-Bench 上，集成 AskChem 后 GPT-5.5 读者可解析 DOI 比例从 88.3% 提升至 100%。

## 全文

Chemistry literature synthesis often requires assembling specific findings scattered across many publications, yet existing literature-search systems primarily return ranked document lists. As a result, scientists and AI agents need to locate relevant information, verify their provenance, and assemble cross-paper answers manually. We present AskChem, a claim-centered infrastructure for cross-paper chemistry search. AskChem changes the unit of retrieval from the paper to the provenance-carrying claim: each paper is converted into atomic, typed claims, each grounded by a source DOI and a verbatim quote or an explicit evidence locator. Over this shared claim store, AskChem exposes complementary structures for search and synthesis: a stabilized faceted taxonomy for hierarchical retrieval and browsing, an evidence graph linking claims through relations, and an exploratory living taxonomy that situates indexed papers under scientific principles. AskChem currently indexes 2.4M claims from 147K papers and provides a web interface, as well as REST, SDK, and MCP access for AI agents. On AskChem-Bench, grounding a GPT-5.5 reader in AskChem yields 100% resolvable DOIs, compared with 88.3% without retrieval, and the highest citation density among five tested systems. AskChem is live at https://askchem.org.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_f2552ca7b9a24c7fc6fcac05