{ "schemaVersion": "1.2", "type": "tutorial-evidence-record", "language": "zh-CN", "generatedAt": "2026-09-28", "publisher": { "id": "https://zenleak.cn/#organization", "name": "ZenLeak GEO 学院", "editorialName": "ZenLeak 编辑部", "contact": "geo@allintwo.cloud" }, "policies": { "sourceRelationship": "Page sources provide general context. Explicit claim evidence includes a source excerpt, locator, relation, review date and limits; unmapped claims do not imply one-to-one source support.", "rankingDisclaimer": "Inclusion in this evidence layer does not guarantee search ranking, indexing, citation, recommendation, or adoption by an AI system." }, "record": { "id": "domestic-ai-citation-benchmark-method", "title": "国内 AI 引用实验怎么做：从固定问题到可复核结果", "url": "https://zenleak.cn/experiments/domestic-ai-citation-benchmark-method/", "path": "/experiments/domestic-ai-citation-benchmark-method/", "evidenceUrl": "https://zenleak.cn/evidence/#domestic-ai-citation-benchmark-method", "representations": { "json": "https://zenleak.cn/evidence/domestic-ai-citation-benchmark-method.json", "markdown": "https://zenleak.cn/evidence/domestic-ai-citation-benchmark-method.md" }, "category": "实验室", "contentType": "experiment", "status": "reviewed", "statusLabel": "已复核", "datePublished": "2026-09-28", "dateModified": "2026-09-28", "reviewedAt": "2026-09-28", "evidenceReviewedAt": "2026-09-28", "claims": [ { "id": "domestic-ai-citation-benchmark-method-claim-1", "text": "一次可复测的国内 AI 引用实验，不是把同一个问题随手问给几个平台，而是预先冻结平台、问题、语言、地区、登录状态、联网模式、页面版本和观察窗口。", "url": "https://zenleak.cn/evidence/#domestic-ai-citation-benchmark-method-claim-1", "articleUrl": "https://zenleak.cn/experiments/domestic-ai-citation-benchmark-method/#domestic-ai-citation-benchmark-method-claim-1", "evidence": [ { "tutorialId": "domestic-ai-citation-benchmark-method", "claimText": "一次可复测的国内 AI 引用实验，不是把同一个问题随手问给几个平台，而是预先冻结平台、问题、语言、地区、登录状态、联网模式、页面版本和观察窗口。", "relation": "first-party", "sourceUrl": "https://zenleak.cn/benchmarks/", "sourceTitle": "ZenLeak 国内 AI 引用基准库", "locator": "实验目的、固定问题集、平台列表与测量计划", "excerpt": "基准库使用固定中文问题和多个国内 AI 平台，分别记录联网、来源、品牌提及、页面引用、内容采用、引用准确和回答完整，并保留观察窗口与限制。", "checkedAt": "2026-09-28", "limits": "这是 ZenLeak 的实验设计与公开记录方法，不是任何 AI 平台的官方测试规范，也不能保证固定条件下得到稳定回答。" } ] }, { "id": "domestic-ai-citation-benchmark-method-claim-2", "text": "ZenLeak 当前基准协议使用 5 个国内 AI 平台、12 个固定中文问题和 3 个观察窗口，共 180 个计划组合；未提交、无法访问和证据不完整的组合必须保留为缺失或未知，不能放进已完成结果的分母。", "url": "https://zenleak.cn/evidence/#domestic-ai-citation-benchmark-method-claim-2", "articleUrl": "https://zenleak.cn/experiments/domestic-ai-citation-benchmark-method/#domestic-ai-citation-benchmark-method-claim-2", "evidence": [ { "tutorialId": "domestic-ai-citation-benchmark-method", "claimText": "ZenLeak 当前基准协议使用 5 个国内 AI 平台、12 个固定中文问题和 3 个观察窗口，共 180 个计划组合；未提交、无法访问和证据不完整的组合必须保留为缺失或未知，不能放进已完成结果的分母。", "relation": "first-party", "sourceUrl": "https://zenleak.cn/benchmarks/", "sourceTitle": "ZenLeak 国内 AI 引用基准库", "locator": "范围、问题集、观察窗口和 null 值规则", "excerpt": "公开协议列出 5 个平台、12 个问题和 3 个窗口；未完成结果保留为 null，不把空值当成 0，也不据此发布总体引用率。", "checkedAt": "2026-09-28", "limits": "180 是计划组合数，不是已经完成的观察数；实际完成度、平台可用性和来源完整度需要按每一轮公开记录核对。" } ] }, { "id": "domestic-ai-citation-benchmark-method-claim-3", "text": "品牌被提及、回答出现来源、来源链接到具体页面、内容被采用、引用支持准确和回答完整是不同维度；其中任意一项为真，都不能自动推出其他维度也为真。", "url": "https://zenleak.cn/evidence/#domestic-ai-citation-benchmark-method-claim-3", "articleUrl": "https://zenleak.cn/experiments/domestic-ai-citation-benchmark-method/#domestic-ai-citation-benchmark-method-claim-3", "evidence": [ { "tutorialId": "domestic-ai-citation-benchmark-method", "claimText": "品牌被提及、回答出现来源、来源链接到具体页面、内容被采用、引用支持准确和回答完整是不同维度；其中任意一项为真，都不能自动推出其他维度也为真。", "relation": "direct", "sourceUrl": "https://arxiv.org/abs/2304.09848", "sourceTitle": "Evaluating Verifiability in Generative Search Engines", "locator": "Abstract：citation recall 与 citation precision", "excerpt": "systems should cite comprehensively (high citation recall; all statements are fully supported by citations) and accurately (high citation precision; every cite supports its associated statement).", "checkedAt": "2026-09-28", "limits": "论文讨论生成式搜索引用评价，不定义 ZenLeak 的品牌提及、内容采用和回答完整字段；本文字段是面向国内平台人工观察的扩展。" } ] }, { "id": "domestic-ai-citation-benchmark-method-claim-4", "text": "高质量记录应能从公开结论回到问题原文、运行条件、完整回答、来源链接、被核查页面版本和审核理由；单张截图或事后摘要不能替代完整证据链。", "url": "https://zenleak.cn/evidence/#domestic-ai-citation-benchmark-method-claim-4", "articleUrl": "https://zenleak.cn/experiments/domestic-ai-citation-benchmark-method/#domestic-ai-citation-benchmark-method-claim-4", "evidence": [ { "tutorialId": "domestic-ai-citation-benchmark-method", "claimText": "高质量记录应能从公开结论回到问题原文、运行条件、完整回答、来源链接、被核查页面版本和审核理由；单张截图或事后摘要不能替代完整证据链。", "relation": "context", "sourceUrl": "https://www.w3.org/TR/prov-overview/", "sourceTitle": "W3C PROV-Overview", "locator": "Provenance information and its use for assessing quality and reliability", "excerpt": "Provenance is information about entities, activities, and people involved in producing a piece of data or thing, which can be used to assess its quality, reliability or trustworthiness.", "checkedAt": "2026-09-28", "limits": "W3C 提供通用 provenance 概念；问题、回答、来源、页面版本和审核字段是 ZenLeak 为 AI 引用观察制定的记录结构，不是 W3C 对 AI 平台的具体要求。" } ] } ], "sources": [ { "label": "ZenLeak 国内 AI 引用基准库", "url": "https://zenleak.cn/benchmarks/", "publisher": "ZenLeak 实验室" }, { "label": "ZenLeak AI 搜索引用证据归档方法", "url": "https://zenleak.cn/content/ai-citation-evidence-record/", "publisher": "ZenLeak 实验室" }, { "label": "Evaluating Verifiability in Generative Search Engines", "url": "https://arxiv.org/abs/2304.09848", "publisher": "arXiv" }, { "label": "Enabling Large Language Models to Generate Text with Citations", "url": "https://aclanthology.org/2023.emnlp-main.398/", "publisher": "ACL Anthology" }, { "label": "PROV-Overview", "url": "https://www.w3.org/TR/prov-overview/", "publisher": "W3C" }, { "label": "Creating helpful, reliable, people-first content", "url": "https://developers.google.com/search/docs/fundamentals/creating-helpful-content", "publisher": "Google Search Central" } ], "prerequisites": [ { "id": "ai-citation-test", "title": "AI 引用测试怎么做：建立 GEO 实验记录", "url": "https://zenleak.cn/experiments/ai-citation-test/" }, { "id": "geo-citation-testing-before-indexed", "title": "搜索引擎收录很少时，如何正确做 GEO 引用测试？", "url": "https://zenleak.cn/content/geo-citation-testing-before-indexed/" }, { "id": "ai-citation-evidence-record", "title": "AI 搜索引用证据怎么归档：从原始回答到可复核记录", "url": "https://zenleak.cn/content/ai-citation-evidence-record/" } ], "related": [ { "id": "ai-citation-test", "title": "AI 引用测试怎么做：建立 GEO 实验记录", "url": "https://zenleak.cn/experiments/ai-citation-test/" }, { "id": "geo-citation-testing-before-indexed", "title": "搜索引擎收录很少时，如何正确做 GEO 引用测试？", "url": "https://zenleak.cn/content/geo-citation-testing-before-indexed/" }, { "id": "ai-citation-evidence-record", "title": "AI 搜索引用证据怎么归档：从原始回答到可复核记录", "url": "https://zenleak.cn/content/ai-citation-evidence-record/" }, { "id": "ai-citation-accuracy", "title": "AI 引用了你的网站，就算 GEO 成功吗？引用准确性的核查方法", "url": "https://zenleak.cn/experiments/ai-citation-accuracy/" }, { "id": "china-ai-search-geo", "title": "国内 AI 搜索如何获取网页信息：DeepSeek、Kimi、豆包、通义千问、智谱对比", "url": "https://zenleak.cn/content/china-ai-search-geo/" }, { "id": "geo-measurement-framework", "title": "如何衡量 GEO 效果：从 AI 引用、搜索可见性到业务转化", "url": "https://zenleak.cn/practice/geo-measurement-framework/" } ] } }