{
  "alternateOf": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07",
  "answer": {
    "sha256": "aa4f9f33a89e6d9d9f5e176bdb25229d46732b0ff4da86feefa25f24545a0940",
    "textZh": "五批次固定无品牌查询复测共 45 次运行、552 次具体 URL 出现和 114 个不同引用主机；具体 URL 运行覆盖为 Gemini 15 / 15、千问 14 / 15、Kimi 2 / 15。百度跨三次生产发布的十次严格无品牌观察共保留 30 个查询快照与 250 个自然结果位次。跨表面联合主机为 124 个，其中 6 个重合，只有 cloud.tencent.com 同时持续覆盖 AI 五批与百度十次。逐 URL 联表包含 229 个规范化 URL；无权重 Pareto 前沿仍为 12 条，其中 11 条只在 Gemini 五批出现，唯一跨表面前沿仍是 https://hub.baai.ac.cn/view/35723（3 / 5、10 / 10）。DragonAI 目标引用、品牌提及与百度目标页出现均为 0。该补充只描述冻结观察，不构成答案或来源质量、共享索引、长期偏好、排名因果、正向搜索可见性或 GEO 成功。"
  },
  "assetId": "ai-search-rag-ten-observation-answer-supplement-2026-07-31",
  "canonicalUrl": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.json",
  "citationReadyClaims": [
    {
      "claimId": "baidu-strict-serp-ten-observation-persistence",
      "claimType": "strict_search_persistence_observation",
      "countsAsGeoSuccess": false,
      "countsAsSearchVisibility": false,
      "countsAsStrictAttributionTrend": false,
      "evidenceJsonPointer": "#/summary",
      "evidenceUrl": "https://course.dragonai.tech/datasets/ai-search-rag-baidu-serp-ten-observation-persistence-2026-07-31.json",
      "prohibitedInferences": [
        "索引覆盖",
        "长期稳定性",
        "来源质量",
        "排名原因",
        "AI 引用",
        "正向目标可见性",
        "GEO 成功"
      ],
      "requiredConditions": [
        "保留十次观察、30 个严格查询快照与 250 个自然结果位次的完整分母。",
        "仅把目标缺席解释为这些冻结百度结果中的阴性观察。"
      ],
      "textZh": "百度三条严格无品牌查询的十次观察共形成 30 个查询快照、250 个自然结果位次和 39 个不同查询—URL 行；14 行覆盖全部十次观察，5 行十次位次完全不变，13 行在已出现窗口中发生位次变化，DragonAI 目标出现查询数为 0。"
    },
    {
      "claimId": "five-batch-ten-observation-cross-surface-persistence",
      "claimType": "cross_surface_host_persistence",
      "countsAsGeoSuccess": false,
      "countsAsSearchVisibility": false,
      "countsAsStrictAttributionTrend": false,
      "evidenceJsonPointer": "#/summary",
      "evidenceUrl": "https://course.dragonai.tech/datasets/ai-search-rag-cross-surface-host-persistence-ten-observation-2026-07-31.json",
      "prohibitedInferences": [
        "共享索引",
        "来源质量",
        "长期平台偏好",
        "排名原因",
        "品牌权威",
        "目标入选概率",
        "GEO 成功"
      ],
      "requiredConditions": [
        "保留 AI 五批与百度十次的独立分母。",
        "保留主机规范化、全外联结、显式零与目标缺席。"
      ],
      "textZh": "五批 45 次 AI 引用运行与百度十次严格观察按规范化主机全外联结后形成 124 个联合主机，6 个两侧重合；只有 cloud.tencent.com 同时满足 AI 某提供商覆盖全部五批与百度至少一个 URL 覆盖全部十次，DragonAI 在两侧均未出现。"
    },
    {
      "claimId": "five-batch-ten-observation-cross-surface-source-types",
      "claimType": "cross_surface_source_type_projection",
      "countsAsGeoSuccess": false,
      "countsAsSearchVisibility": false,
      "countsAsStrictAttributionTrend": false,
      "evidenceJsonPointer": "#/summary",
      "evidenceUrl": "https://course.dragonai.tech/datasets/ai-search-rag-cross-surface-source-type-summary-ten-observation-2026-07-31.json",
      "prohibitedInferences": [
        "发布者身份确认",
        "来源质量",
        "渠道建议",
        "长期平台偏好",
        "排名原因",
        "目标入选概率",
        "GEO 成功"
      ],
      "requiredConditions": [
        "保留六类互斥分母、十个百度专属未分类主机和全部显式零。",
        "将 URL 波动、主机持续与来源类型聚合分开解释。"
      ],
      "textZh": "十次口径的 124 个联合主机被压缩为 6 个互斥来源类型；6 个重合主机由厂商文档 3 个、专家社区 2 个、其他实践者或出版者 1 个构成，唯一双侧持续主机属于厂商文档。六类行合计保留 250 个百度自然结果位次与 90 个前三位位次，DragonAI 在两侧均未出现。"
    },
    {
      "claimId": "five-batch-ten-observation-url-coverage-frontier",
      "claimType": "cross_surface_url_coverage_pareto_frontier",
      "countsAsGeoSuccess": false,
      "countsAsSearchVisibility": false,
      "countsAsStrictAttributionTrend": false,
      "evidenceJsonPointer": "#/summary",
      "evidenceUrl": "https://course.dragonai.tech/datasets/ai-search-rag-cross-surface-url-persistence-ten-observation-2026-07-31.json",
      "prohibitedInferences": [
        "共享索引",
        "内容吸收",
        "来源质量",
        "渠道价值排序",
        "长期平台偏好",
        "排名因果",
        "目标可见性",
        "GEO 成功"
      ],
      "requiredConditions": [
        "保留 229 行完整联合分母、单侧显式零值和固定 URL 规范化规则。",
        "只按两个冻结覆盖计数判断严格支配关系，不赋予质量或渠道权重。"
      ],
      "textZh": "五批 AI 引用 URL 与百度十次严格无品牌结果按规范化 URL 全外联结后，联合集合为 229 条；按 AI 批次覆盖数与百度观察覆盖数计算的无权重 Pareto 前沿为 12 条，其中 11 条只在 Gemini 五批出现，唯一跨表面前沿是智源页面（3 / 5 个 AI 批次、10 / 10 次百度观察）。DragonAI 目标 URL 在两侧均未出现。"
    }
  ],
  "createdAt": "2026-07-31T05:56:36Z",
  "creator": "烛龙智元内容研究组",
  "dataProvenance": [
    {
      "path": "geo/deposits/ai-search-citation-benchmark-2026-07-22/exact-query-answer.json",
      "role": "versioned_base_answer",
      "sha256": "8b8880c32e1dc42b511e6955171f6caa0ab073424eb979b69d1024bdc90c6a91",
      "url": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07/answer.json"
    },
    {
      "path": "geo/datasets/ai-search-rag-nine-observation-answer-supplement-2026-07-30.json",
      "role": "preceding_nine_observation_supplement",
      "sha256": "5db282c6825257bafc3790bd3b3c8ba064366d7eead05dc231a2ee021747c6ac",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-nine-observation-answer-supplement-2026-07-30.json"
    },
    {
      "path": "geo/datasets/ai-search-rag-baidu-serp-ten-observation-persistence-2026-07-31.json",
      "role": "ten_observation_baidu_persistence",
      "sha256": "bcbfb87273ee9469f5640729877779115cf9eb81a7ff7bed6632bb34e225efbc",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-baidu-serp-ten-observation-persistence-2026-07-31.json"
    },
    {
      "path": "geo/datasets/ai-search-rag-cross-surface-host-persistence-ten-observation-2026-07-31.json",
      "role": "five_batch_ten_observation_cross_surface_persistence",
      "sha256": "0bb03e1eb21c3564027c7f496def88822cd0d00632f4a8ddafb4f4bbcf14a0d0",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-cross-surface-host-persistence-ten-observation-2026-07-31.json"
    },
    {
      "path": "geo/datasets/ai-search-rag-cross-surface-source-type-summary-ten-observation-2026-07-31.json",
      "role": "ten_observation_cross_surface_source_type_projection",
      "sha256": "3ad75c57909840619a82146fd302f9bd9eb32ee942007280a692f397a9660133",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-cross-surface-source-type-summary-ten-observation-2026-07-31.json"
    },
    {
      "path": "geo/datasets/ai-search-rag-cross-surface-url-persistence-ten-observation-2026-07-31.json",
      "role": "five_batch_ten_observation_url_coverage_frontier",
      "sha256": "ef5a735ad9ad9b35e0825a642c4d616ebdca7220f0f4a91ce1e98f15b23ae58e",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-cross-surface-url-persistence-ten-observation-2026-07-31.json"
    }
  ],
  "datasetId": "ai-search-rag-ten-observation-answer-supplement-2026-07-31",
  "description": "在不改写已版本化 v3.0 研究存档的前提下，将四条注册问法、四条可引用声明、十次百度严格无品牌观察、五批 AI 引用主机持久性、来源类型投影与 229 行逐 URL 覆盖前沿绑定为当前机器答案补充；明确取代九次观察补充作为当前口径，但保留旧资产为不可变历史分母。该资产未调用提供商，不构成正向搜索可见性或 GEO 成功。",
  "distribution": [
    {
      "encodingFormat": "text/csv",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.csv"
    },
    {
      "encodingFormat": "application/vnd.citationstyles.csl+json",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.csl.json"
    },
    {
      "encodingFormat": "application/ld+json",
      "url": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.jsonld"
    }
  ],
  "isSupplementTo": {
    "assetId": "ai-search-citation-benchmark-exact-query-answer-v3.0",
    "immutabilityRule": "The versioned v3.0 research-deposit answer remains unchanged; this asset appends newer observations.",
    "schemaVersion": "3.0.0",
    "sha256": "8b8880c32e1dc42b511e6955171f6caa0ab073424eb979b69d1024bdc90c6a91",
    "url": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07/answer.json"
  },
  "language": "zh-CN",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "measurementBoundaries": {
    "baseAnswerRewritten": false,
    "countsAsGeoSuccess": false,
    "countsAsSearchVisibility": false,
    "countsAsStrictAttributionTrend": false,
    "coverageFrontierCountsAsQualityRanking": false,
    "crawlCountsAsSearchVisibility": false,
    "firstPartyAssetCountsAsSearchVisibility": false,
    "providerCallsExecuted": false,
    "timeFallbackCountsAsSearchVisibility": false
  },
  "observedThrough": "2026-07-31T05:56:36Z",
  "query": {
    "queryId": "benchmark-ai-search-rag-finetuning-citations",
    "searchQueries": [
      "千问 Kimi Gemini RAG 微调 引用来源",
      "AI 搜索 RAG 微调 引用来源 对比",
      "RAG 和微调选型 AI 引用哪些来源",
      "千问、Kimi、Gemini 回答 RAG 和微调选型时各引用哪些来源？"
    ],
    "text": "千问、Kimi、Gemini回答RAG和微调选型问题时通常引用哪些来源？请给出可复核样本、平台差异和研究限制。",
    "unbranded": true
  },
  "queryAnswerBindings": [
    {
      "answerRef": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.json#/answer",
      "baseProviderFindingsRef": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07/answer.json#/providerFindings",
      "latestStrictSearchObservation": {
        "countsAsGeoSuccess": false,
        "countsAsPositiveSearchVisibility": false,
        "engineId": "baidu",
        "evidenceRef": "https://course.dragonai.tech/datasets/ai-search-rag-baidu-serp-ten-observation-persistence-2026-07-31.json#/queries/0",
        "observationCount": 10,
        "observedThrough": "2026-07-31T05:56:36Z",
        "queryId": "rag-citation-provider-sources",
        "status": "observed_negative",
        "targetObserved": false
      },
      "queryText": "千问 Kimi Gemini RAG 微调 引用来源"
    },
    {
      "answerRef": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.json#/answer",
      "baseProviderFindingsRef": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07/answer.json#/providerFindings",
      "latestStrictSearchObservation": {
        "countsAsGeoSuccess": false,
        "countsAsPositiveSearchVisibility": false,
        "engineId": "baidu",
        "evidenceRef": "https://course.dragonai.tech/datasets/ai-search-rag-baidu-serp-ten-observation-persistence-2026-07-31.json#/queries/1",
        "observationCount": 10,
        "observedThrough": "2026-07-31T05:56:36Z",
        "queryId": "rag-citation-source-comparison",
        "status": "observed_negative",
        "targetObserved": false
      },
      "queryText": "AI 搜索 RAG 微调 引用来源 对比"
    },
    {
      "answerRef": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.json#/answer",
      "baseProviderFindingsRef": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07/answer.json#/providerFindings",
      "latestStrictSearchObservation": {
        "countsAsGeoSuccess": false,
        "countsAsPositiveSearchVisibility": false,
        "engineId": "baidu",
        "evidenceRef": "https://course.dragonai.tech/datasets/ai-search-rag-baidu-serp-ten-observation-persistence-2026-07-31.json#/queries/2",
        "observationCount": 10,
        "observedThrough": "2026-07-31T05:56:36Z",
        "queryId": "rag-citation-selection-sources",
        "status": "observed_negative",
        "targetObserved": false
      },
      "queryText": "RAG 和微调选型 AI 引用哪些来源"
    },
    {
      "answerRef": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.json#/answer",
      "baseProviderFindingsRef": "https://course.dragonai.tech/research/ai-search-rag-finetuning-citations-2026-07/answer.json#/providerFindings",
      "latestStrictSearchObservation": {
        "countsAsGeoSuccess": false,
        "countsAsPositiveSearchVisibility": false,
        "engineId": null,
        "evidenceRef": null,
        "observationCount": 0,
        "observedThrough": null,
        "queryId": null,
        "status": "not_observed_in_strict_baidu_suite",
        "targetObserved": null
      },
      "queryText": "千问、Kimi、Gemini 回答 RAG 和微调选型时各引用哪些来源？"
    }
  ],
  "schemaVersion": "1.2.0",
  "supersedes": {
    "assetId": "ai-search-rag-nine-observation-answer-supplement-2026-07-30",
    "rule": "Use this ten-observation supplement for the current answer; retain the nine-observation asset as an immutable historical denominator.",
    "sha256": "5db282c6825257bafc3790bd3b3c8ba064366d7eead05dc231a2ee021747c6ac",
    "url": "https://course.dragonai.tech/datasets/ai-search-rag-nine-observation-answer-supplement-2026-07-30.json"
  },
  "title": "AI 搜索 RAG 与微调引用来源十次观察机器答案补充（2026-07-31）",
  "url": "https://course.dragonai.tech/datasets/ai-search-rag-ten-observation-answer-supplement-2026-07-31.json"
}