{"id":4366,"date":"2025-11-17T10:04:18","date_gmt":"2025-11-17T02:04:18","guid":{"rendered":"https:\/\/www.15zhi.net\/blog\/?p=4366"},"modified":"2025-11-17T10:42:01","modified_gmt":"2025-11-17T02:42:01","slug":"202511%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-past-meets-present-creating-historical-analogy-with-large-language-models","status":"publish","type":"post","link":"https:\/\/www.15zhi.net\/blog\/202511%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-past-meets-present-creating-historical-analogy-with-large-language-models\/","title":{"rendered":"202511\u8bba\u6587\u7814\u8bfb-Past Meets Present: Creating Historical Analogy with Large Language Models"},"content":{"rendered":"\n

\u6765\u6e90\uff1aACL
\u4f5c\u8005\uff1aNianqi Li\u7b49
\u5355\u4f4d\uff1aShanghai Key Laboratory of Data Science\uff0cFudan University.
\u53d1\u8868\u65f6\u95f4\uff1a2025 \u5e74 8\u6708<\/p>\n\n\n\n

\u4e00\u3001\u7814\u7a76\u80cc\u666f<\/h2>\n\n\n\n

\u5386\u53f2\u7c7b\u6bd4\u7684\u91cd\u8981\u6027
\u5386\u53f2\u7c7b\u6bd4\u901a\u8fc7\u5c06\u5f53\u4ee3\u964c\u751f\u4e8b\u4ef6\u4e0e\u5df2\u77e5\u5386\u53f2\u4e8b\u4ef6\u8fdb\u884c\u6bd4\u8f83\uff0c\u5e2e\u52a9\u4eba\u4eec\u7406\u89e3\u73b0\u5b9e\u3001\u8f85\u52a9\u51b3\u7b56\u3002
\u5e7f\u6cdb\u5e94\u7528\u4e8e\u653f\u7b56\u5236\u5b9a\u3001\u5371\u673a\u5e94\u5bf9\uff08\u5982\u75281918\u5927\u6d41\u611f\u7c7b\u6bd4\u65b0\u51a0\u75ab\u60c5\uff09\u3002<\/p>\n\n\n\n

\u4eba\u7c7b\u5728\u4f7f\u7528\u5386\u53f2\u7c7b\u6bd4\u65f6\u7684\u5c40\u9650\u6027
\u5bb9\u6613\u4f9d\u8d56\u8868\u9762\u76f8\u4f3c\u6027\uff1b
\u5e38\u9009\u62e9\u6700\u5148\u60f3\u5230\u7684\u7c7b\u6bd4\uff0c\u7f3a\u4e4f\u7cfb\u7edf\u5206\u6790\uff1b
\u9700\u8981\u5e7f\u535a\u7684\u5386\u53f2\u77e5\u8bc6\uff0c\u95e8\u69db\u9ad8\u3002<\/p>\n\n\n\n

\u8fd9\u7bc7\u6587\u7ae0\u63d0\u51fa\u201c\u5386\u53f2\u7c7b\u6bd4\u83b7\u53d6\u201d\u4efb\u52a1\uff1a\u7ed9\u5b9a\u4e00\u4e2a\u5f53\u4ee3\u4e8b\u4ef6\u53ca\u5176\u63cf\u8ff0\uff0c\u81ea\u52a8\u751f\u6210\u4e00\u4e2a\u5728\u591a\u7ef4\u5ea6\u4e0a\u76f8\u4f3c\u7684\u5386\u53f2\u4e8b\u4ef6\u3002<\/p>\n\n\n\n

\u4e8c\u3001\u4e3b\u8981\u6846\u67b6<\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n
    \n
  1. \u4efb\u52a1\u5b9a\u4e49\uff1a\u5386\u53f2\u7c7b\u6bd4\u83b7\u53d6\u8f93\u5165\uff1a\u4e00\u4e2a\u5f53\u4ee3\u4e8b\u4ef6 E \u53ca\u5176\u6587\u672c\u63cf\u8ff0 D
    \u4f8b\u5982\uff1a\u201c2020\u5e74\u7f8e\u56fd\u56fd\u4f1a\u5c71\u9a9a\u4e71\u201d\uff09
    \u8f93\u51fa\uff1a\u4e00\u4e2a\u5386\u53f2\u4e0a\u771f\u5b9e\u5b58\u5728\u7684\u4e8b\u4ef6 \u5728\u591a\u4e2a\u7ef4\u5ea6\u4e0a\u4e0e E\u6784\u6210\u6709\u6548\u7c7b\u6bd4<\/li>\n<\/ol>\n\n\n\n

    2. \u7c7b\u6bd4\u7ef4\u5ea6\uff08\u57fa\u4e8e\u53f2\u5b66\u7406\u8bba\uff09\uff1a
    Topic\uff08\u4e3b\u9898\uff09\uff1a\u4e8b\u4ef6\u7684\u6838\u5fc3\u8bae\u9898\uff08\u5982\u201c\u653f\u6cbb\u66b4\u529b\u201d\uff09
    Background\uff08\u80cc\u666f\uff09\uff1a\u53d1\u751f\u7684\u793e\u4f1a\/\u653f\u6cbb\/\u7ecf\u6d4e\u6761\u4ef6
    Process\uff08\u8fc7\u7a0b\uff09\uff1a\u4e8b\u4ef6\u5982\u4f55\u53d1\u5c55\u3001\u5173\u952e\u8282\u70b9
    Result\uff08\u7ed3\u679c\uff09\uff1a\u77ed\u671f\/\u957f\u671f\u5f71\u54cd\u3001\u540e\u679c
    \u76ee\u6807\u4e0d\u4ec5\u662f\u201c\u770b\u8d77\u6765\u50cf\u201d\uff0c\u800c\u662f\u7ed3\u6784\u76f8\u4f3c\u3001\u903b\u8f91\u53ef\u6bd4\u3002<\/p>\n\n\n\n

    \u8303\u5f0f<\/td>\u6838\u5fc3\u601d\u60f3<\/td>\u4f18\u70b9<\/td>\u7f3a\u70b9<\/td><\/tr>
    \u6570\u636e\u96c6\u68c0\u7d22\u6cd5<\/td>\u4ece\u56fa\u5b9a\u5386\u53f2\u4e8b\u4ef6\u5e93\u4e2d\u67e5\u627e\u6700\u76f8\u4f3c\u9879<\/td>\u7ed3\u679c\u771f\u5b9e\u3001\u53ef\u63a7<\/td>\u8986\u76d6\u6709\u9650\u3001\u66f4\u65b0\u56f0\u96be<\/td><\/tr>
    \u81ea\u7531\u751f\u6210\u6cd5<\/td>\u5229\u7528LLM\u5185\u90e8\u77e5\u8bc6\u76f4\u63a5\u751f\u6210\u7c7b\u6bd4<\/td>\u7075\u6d3b\u3001\u8986\u76d6\u9762\u5e7f<\/td>\u6613\u4ea7\u751f\u5e7b\u89c9\u3001\u523b\u677f\u5370\u8c61<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

    3. \u4f7f\u7528 Google Arts & Culture \u4e2d\u7684658\u4e2a\u5386\u53f2\u4e8b\u4ef6\u4f5c\u4e3a\u5019\u9009\u6c60\u3002
    (1) \u76f4\u63a5\u68c0\u7d22\uff08Direct Retrieval\uff09
    \u6b65\u9aa4\uff1a
    \u7528 text-embedding-3-small \u5bf9\u8f93\u5165\u4e8b\u4ef6\u63cf\u8ff0\u548c\u6240\u6709\u5019\u9009\u4e8b\u4ef6\u63cf\u8ff0\u7f16\u7801\uff1b
    \u8ba1\u7b97\u4f59\u5f26\u76f8\u4f3c\u5ea6\uff1b
    \u8fd4\u56de\u76f8\u4f3c\u5ea6\u6700\u9ad8\u7684\u5386\u53f2\u4e8b\u4ef6\u3002
    \u7279\u70b9\uff1a\u7b80\u5355\u9ad8\u6548\uff0c\u4f46\u4ec5\u4f9d\u8d56\u8868\u9762\u8bed\u4e49\u5339\u914d\u3002
    (2) \u4e24\u9636\u6bb5\u68c0\u7d22\uff08Two-stage Retrieval\uff09
    \u6b65\u9aa4\uff1a
    \u5148\u7528\u5d4c\u5165\u53ec\u56de Top-10 \u5019\u9009\uff1b
    \u5c06\u8fd910\u4e2a\u4e8b\u4ef6\u7684\u63cf\u8ff0\u8f93\u5165LLM\uff0c\u63d0\u793a\u5176\u9009\u62e9\u201c\u6700\u5408\u9002\u7684\u7c7b\u6bd4\u201d\u3002
    \u4f18\u52bf\uff1a\u5f15\u5165LLM\u7684\u63a8\u7406\u80fd\u529b\uff0c\u63d0\u5347\u76f8\u5173\u6027\u3002<\/p>\n\n\n\n

    \"\"<\/figure>\n\n\n\n

    4. \u5229\u7528LLM\u53c2\u6570\u4e2d\u9690\u542b\u7684\u5386\u53f2\u77e5\u8bc6\uff0c\u65e0\u9700\u4f9d\u8d56\u5916\u90e8\u6570\u636e\u5e93\u3002<\/p>\n\n\n\n

    (1) \u76f4\u63a5\u751f\u6210\uff08Direct Generation\uff09
    \u201cGiven the event \u2018X\u2019, generate a historically analogous event.\u201d
    \u95ee\u9898\uff1a\u6613\u53d7\u8bad\u7ec3\u6570\u636e\u504f\u89c1\u5f71\u54cd\uff0c\u53ef\u80fd\u751f\u6210\u865a\u6784\u4e8b\u4ef6\uff08\u5982\u201c19\u4e16\u7eaa\u6cd5\u56fdAI\u9769\u547d\u201d\uff09\u3002<\/p>\n\n\n\n

    (2) \u4e24\u9636\u6bb5\u751f\u6210\uff08Two-stage Generation\uff09
    LLM\u751f\u621010\u4e2a\u5019\u9009\u5386\u53f2\u4e8b\u4ef6\uff1b
    \u901a\u8fc7Wikipedia API\u9a8c\u8bc1\u6bcf\u4e2a\u5019\u9009\u662f\u5426\u771f\u5b9e\u5b58\u5728\uff1b
    \u5bf9\u771f\u5b9e\u5019\u9009\uff0c\u83b7\u53d6\u5176\u7ef4\u57fa\u63cf\u8ff0\uff1b
    LLM\u6bd4\u8f83\u8f93\u5165\u4e8b\u4ef6\u4e0e\u5019\u9009\u63cf\u8ff0\uff0c\u9009\u51fa\u6700\u4f73\u7c7b\u6bd4\u3002<\/p>\n\n\n\n

    \"\"<\/figure>\n\n\n\n

    (3)\u5e26\u6458\u8981\u751f\u6210\uff08Generation with Summarizing\uff09
    \u521b\u65b0\u70b9\uff1a\u5148\u5c06\u8f93\u5165\u4e8b\u4ef6\u548c\u5019\u9009\u4e8b\u4ef6\u5206\u522b\u6309\u56db\u4e2a\u7ef4\u5ea6\uff08Topic, Background, Process, Result\uff09\u7ed3\u6784\u5316\u6458\u8981\uff08\u7531GPT-4\u5b8c\u6210\uff09\uff0c\u518d\u57fa\u4e8e\u8fd9\u4e9b\u6458\u8981\u8fdb\u884c\u6bd4\u8f83\u548c\u9009\u62e9\u3002
    \u6548\u679c\uff1a\u4fe1\u606f\u66f4\u6d53\u7f29\u3001\u5bf9\u6bd4\u66f4\u805a\u7126\uff0c\u63d0\u5347\u7c7b\u6bd4\u8d28\u91cf\u3002<\/p>\n\n\n\n

    (4) \u81ea\u53cd\u601d\u6846\u67b6\uff08Self-reflection Framework\uff09 \uff1a\u8fed\u4ee3\u4f18\u5316\u673a\u5236\uff0c\u5305\u542b\u4e24\u4e2aLLM\u6a21\u5757\uff1a<\/p>\n\n\n\n

    \u6a21\u5757<\/td>\u529f\u80fd<\/td><\/tr>
    Candidate Generator\uff08\u5019\u9009\u751f\u6210\u5668\uff09<\/td>\u57fa\u4e8e\u56db\u7ef4\u6458\u8981\uff0c\u6bcf\u6b21\u751f\u62105\u4e2a\u5019\u9009\u5386\u53f2\u4e8b\u4ef6<\/td><\/tr>
    Answer Reflector\uff08\u7b54\u6848\u53cd\u601d\u5668\uff09<\/td>\u8bc4\u4f30\u5019\u9009\uff1a\u2022 \u82e5\u6709\u5408\u9002\u7c7b\u6bd4 \u2192 \u8f93\u51fa\u7ed3\u679c\u2022 \u82e5\u65e0 \u2192 \u7ed9\u51fa\u53cd\u9988\uff08\u5982\u201c\u907f\u514d\u805a\u7126\u540c\u4e00\u56fd\u5bb6\u201d\uff09\uff0c\u8981\u6c42\u91cd\u65b0\u751f\u6210<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

    \u4e09\u3001\u5b9e\u9a8c<\/h2>\n\n\n\n

    \u6570\u636e\u96c6\u6784\u5efa\uff1a
    Popular Analogy\uff0820\u4e2a\uff09\uff1a\u516c\u4f17\u719f\u77e5\u3001\u6709\u6807\u51c6\u7b54\u6848\u7684\u7c7b\u6bd4\uff08\u5982\u65b0\u51a0 vs \u897f\u73ed\u7259\u6d41\u611f\uff09\u3002
    General Analogy\uff08160\u4e2a\uff09\uff1a\u4eceGoogle Arts & Culture\u6536\u96c6\uff0c\u6db5\u76d6\u6218\u4e89\u3001\u653f\u6cbb\u3001\u6587\u5316\u3001\u7ecf\u6d4e\u56db\u5927\u4e3b\u9898\uff0c\u65e0\u6807\u51c6\u7b54\u6848\u3002
    \u6a21\u578b\u9009\u62e9\uff1a
    \u95ed\u6e90\u6a21\u578b\uff1agpt-3.5-turbo-0125\uff08OpenAI\uff09
    \u5f00\u6e90\u6a21\u578b\uff1aLlama3.1-8B-Instruct\uff08Meta\uff09
    \u6e29\u5ea6\uff08temperature\uff09\u7edf\u4e00\u8bbe\u4e3a 0.1\uff0c\u786e\u4fdd\u8f93\u51fa\u7a33\u5b9a\u6027<\/p>\n\n\n\n

    \u8bc4\u4f30\u65b9\u5f0f\uff1a
    \u4eba\u5de5\u8bc4\u4f30\uff1a\u5386\u53f2\u4e13\u4e1a\u5b66\u751f\u5bf9\u751f\u6210\u7ed3\u679c\u6392\u5e8f\uff08Fleiss\u2019s \u03ba=0.97\uff0c\u9ad8\u4e00\u81f4\u6027\uff09\u3002
    \u81ea\u52a8\u8bc4\u4f30\uff1a\u63d0\u51fa \u591a\u7ef4\u76f8\u4f3c\u5ea6\u6307\u6807 MDS\uff08Multi-dimensional Similarity\uff09\uff1a
    \u7ef4\u5ea6\uff1a\u4e3b\u9898\uff08Topic\uff09\u3001\u80cc\u666f\uff08Background\uff09\u3001\u8fc7\u7a0b\uff08Process\uff09\u3001\u7ed3\u679c\uff08Result\uff09\uff1b
    \u7ed3\u5408 \u62bd\u8c61\u76f8\u4f3c\u5ea6\uff08GPT-4\u6253\u5206\uff09\u4e0e \u5b57\u9762\u76f8\u4f3c\u5ea6\uff08Jaccard\uff0c\u53bb\u91cd\u8bcd\uff09\uff1b<\/p>\n\n\n\n

    \"\"<\/figure>\n\n\n\n

    \u5b9e\u9a8c\u7ed3\u679c\uff1a
    <\/p>\n\n\n\n

    \"\"<\/figure>\n\n\n\n

    \u4eba\u5de5\u8bc4\u4f30\u7ed3\u679c
    Self-reflection \u83b7\u5f97\u6700\u9ad8\u5e73\u5747\u6392\u540d\u5206\u548c\u6700\u9ad8\u201c\u88ab\u8bc4\u4e3a\u6700\u4f73\u201d\u6982\u7387\uff1b
    Direct Generation \u5728 Popular \u4e0a\u8868\u73b0\u5c1a\u53ef\uff0c\u4f46\u5728 General \u4e0a\u660e\u663e\u4e0b\u964d\uff1b
    \u4eba\u5de5\u8bc4\u4ef7\u8d8b\u52bf\u4e0e\u81ea\u52a8\u8bc4\u4f30 \u9ad8\u5ea6\u4e00\u81f4\uff0c\u9a8c\u8bc1 MDS \u53ef\u9760\u6027\u3002<\/p>\n\n\n\n

    \"\"<\/figure>\n\n\n\n

    \u56db\u3001\u603b\u7ed3\u4e0e\u7efc\u5408\u5bf9\u9f50\u601d\u8003<\/h2>\n\n\n\n

    \uff08\u4e00\uff09\u8bba\u6587\u6838\u5fc3\u5185\u5bb9
    \u63d0\u51fa\u201c\u5386\u53f2\u7c7b\u6bd4\u83b7\u53d6\u201d\u4efb\u52a1\uff0c\u7ed9\u5b9a\u5f53\u4ee3\u4e8b\u4ef6\uff0c\u81ea\u52a8\u751f\u6210\u7ed3\u6784\u76f8\u4f3c\u7684\u5386\u53f2\u4e8b\u4ef6\uff08\u4ece\u4e3b\u9898\u3001\u80cc\u666f\u3001\u8fc7\u7a0b\u3001\u7ed3\u679c\u56db\u7ef4\u5ea6\u7c7b\u6bd4\uff09\u3002\u5176\u6838\u5fc3\u8d21\u732e\u5305\u62ec\uff1a<\/p>\n\n\n\n


    \u6784\u5efa\u9996\u4e2a\u5386\u53f2\u7c7b\u6bd4\u8bc4\u6d4b\u57fa\u51c6\uff08\u542b20\u4e2a\u6d41\u884c\u7c7b\u6bd4 + 160\u4e2a\u901a\u7528\u7c7b\u6bd4\uff09\uff0c\u5e76\u8bbe\u8ba1\u591a\u7ef4\u81ea\u52a8\u8bc4\u4f30\u6307\u6807 MDS\uff0c\u4e0e\u4eba\u7c7b\u5224\u65ad\u9ad8\u5ea6\u4e00\u81f4\uff08r > 0.72\uff09\u3002
    \u8fd9\u7bc7\u6587\u7ae0\u63d0\u51fa\u4e00\u79cd\u57fa\u4e8e\u53cc\u91cd\u63a8\u7406\u5faa\u73af\u7684\u81ea\u53cd\u601d\u6846\u67b6\uff08Self-reflection Framework\uff09\uff0c\u901a\u8fc7\u534f\u540c\u8c03\u7528LLM\u7684\u751f\u6210\u5f0f\u63a8\u7406\u4e0e\u6279\u5224\u6027\u8bc4\u4f30\u80fd\u529b\uff0c\u5b9e\u73b0\u5bf9\u5386\u53f2\u7c7b\u6bd4\u7684\u8fed\u4ee3\u4f18\u5316\u3002<\/p>\n\n\n\n

    \uff08\u4e8c\uff09\u4e09\u7ef4\u5ea6\u5bf9\u9f50\u601d\u8003
    1.0 \u6280\u672f\u521b\u65b0-\u903b\u8f91\u601d\u7ef4\u63a8\u7406\u6846\u67b6
    \u8fd9\u7bc7\u8bba\u6587\u63d0\u51fa\u7684\u591a\u7ef4\u7ed3\u6784\u5316\u6bd4\u8f83\u673a\u5236\uff08\u4e3b\u9898\/\u80cc\u666f\/\u8fc7\u7a0b\/\u7ed3\u679c\uff09\u542f\u53d1\u98df\u54c1\u9886\u57df\u7684\u63a8\u7406\u7ef4\u5ea6\uff08\u8425\u517b\u6210\u5206\u3001\u5065\u5eb7\u76ee\u6807\u9002\u914d\u3001\u884c\u4e3a\u53cb\u597d\u6027\u3001\u7efc\u5408\u4f18\u9009\uff09\uff0c\u786e\u4fdd\u6bd4\u8f83\u903b\u8f91\u7cfb\u7edf\u4e14\u53ef\u89e3\u91ca\u3002\u540c\u65f6\u53c2\u8003\u5176\u81ea\u53cd\u601d\u6846\u67b6\u901a\u8fc7\u201c\u751f\u6210\u2013\u6279\u5224\u2013\u4fee\u6b63\u201d\u95ed\u73af\uff0c\u4ece\u800c\u63d0\u5347\u63a8\u8350\u53ef\u9760\u6027\u3002
    2.0 \u6280\u672f\u76ee\u6807-\u4e13\u4e1a\u624b\u518c\u516c\u4f17\u670d\u52a1
    \u4e3b\u8981\u662f\u5728\u98df\u54c1\u8fd9\u4e2a\u9886\u57df\uff0c\u5e2e\u52a9\u7528\u6237\u964d\u4f4e\u51b3\u7b56\u6210\u672c\uff0c\u7ed3\u5408\u8425\u517b\u6210\u5206\u542b\u91cf\uff0c\u7528\u6237\u4e2a\u6027\u5316\u559c\u597d\uff0c\u81ea\u52a8\u5e2e\u7528\u6237\u9009\u51fa\u66f4\u9002\u5408\u4e14\u66f4\u5065\u5eb7\u7684\u98df\u54c1\u3002
    3.0 \u573a\u666f\u529f\u80fd-\u98df\u517b\u901a
    \u5728pk\u6bd4\u9009\u8fd9\u4e2a\u90e8\u5206\u8fdb\u884c\u5e94\u7528\uff0c\u9488\u5bf9\u7528\u6237\u9009\u4e2d\u7684\u98df\u54c1\uff0c\u5728\u7cfb\u7edf\u4e2d\u81ea\u52a8\u63a8\u51fa\u7ed3\u5408\u7528\u6237\u4e2a\u6027\u5316\u7684\u66f4\u5065\u5eb7\u4ea7\u54c1\uff0c\u5e76\u5c55\u793a\u51fa\u5bf9\u6bd4\u6027\u89e3\u91ca\u8bf4\u660e\u63d0\u5347\u63a8\u8350\u6548\u679c\u3002<\/p>\n\n\n\n

    <\/p>\n","protected":false},"excerpt":{"rendered":"

    \u6765\u6e90\uff1aACL\u4f5c\u8005\uff1aNianqi Li\u7b49\u5355\u4f4d\uff1aShanghai Key Laboratory of Data S […]<\/p>\n","protected":false},"author":60,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-4366","post","type-post","status-publish","format-standard","hentry","category-events"],"_links":{"self":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/4366","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/users\/60"}],"replies":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/comments?post=4366"}],"version-history":[{"count":2,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/4366\/revisions"}],"predecessor-version":[{"id":4374,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/4366\/revisions\/4374"}],"wp:attachment":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/media?parent=4366"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/categories?post=4366"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/tags?post=4366"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}