{"id":3688,"date":"2025-07-17T16:12:55","date_gmt":"2025-07-17T08:12:55","guid":{"rendered":"https:\/\/www.15zhi.net\/blog\/?p=3688"},"modified":"2025-07-17T16:12:55","modified_gmt":"2025-07-17T08:12:55","slug":"202507%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-scholarsearch-benchmarking-scholar-searching-ability-of-llms","status":"publish","type":"post","link":"https:\/\/www.15zhi.net\/blog\/202507%e8%ae%ba%e6%96%87%e7%a0%94%e8%af%bb-scholarsearch-benchmarking-scholar-searching-ability-of-llms\/","title":{"rendered":"202507\u8bba\u6587\u7814\u8bfb-ScholarSearch: Benchmarking Scholar Searching Ability of LLMs"},"content":{"rendered":"\n

\u5355\u4f4d\uff1aArxiv<\/p>\n\n\n\n

\u4f5c\u8005\uff1aJunting Zhou*, Wang Li*, Yiyan Liao*,Nengyuan Zhang, Tingjia Miao, Zhihui Qi, Yuhan Wu\u2020, Tong Yang\u2020<\/p>\n\n\n\n

\u5b66\u6821\uff1a\u5317\u4eac\u5927\u5b66<\/p>\n\n\n\n

\u65f6\u95f4\uff1a2025\u5e746\u670820\u65e5<\/p>\n\n\n\n

\u4e00\u3001\u80cc\u666f<\/p>\n\n\n\n

\u5317\u4eac\u5927\u5b66DS-Lab\u53d1\u5e03ScholarSearch\uff0c\u8fd9\u662f\u9996\u4e2a\u4e13\u95e8\u7528\u4e8e\u8bc4\u4f30\u5927\u8bed\u8a00\u6a21\u578b\u5728\u5b66\u672f\u7814\u7a76\u4e2d\u590d\u6742\u4fe1\u606f\u68c0\u7d22\u80fd\u529b\u7684\u6570\u636e\u96c6\uff0c\u5305\u542b223\u9053\u9ad8\u96be\u5ea6\u7684\u5b66\u672f\u68c0\u7d22\u9898\u76ee\u53ca\u5176\u7b54\u6848\u3002<\/strong><\/p>\n\n\n\n

\u5b83\u5bf9\u5177\u5907\u8054\u7f51\u641c\u7d22\u80fd\u529b\u7684\u4ee3\u8868\u6027\u6a21\u578b\u53ca\u7eaf\u63a8\u7406\u6a21\u578b\u8fdb\u884c\u4e86\u8bc4\u4f30\uff0c\u7ed3\u679c\u663e\u793a\uff0c\u9876\u5c16\u7684\u7eaf\u63a8\u7406\u6a21\u578b\uff0c\u5982GPT-4.1\u3001DeepSeek-R1\uff0c\u5728\u5904\u7406\u8fd9\u4e9b\u95ee\u9898\u65f6\u51c6\u786e\u7387\u666e\u904d\u4f4e\u4e8e9%\u3002<\/p>\n\n\n\n

\u5177\u5907\u641c\u7d22\u529f\u80fd\u7684\u6a21\u578b\uff0c\u76f8\u8f83\u4e8e\u5176\u65e0\u641c\u7d22\u80fd\u529b\u7684\u7248\u672c\uff0c\u51c6\u786e\u7387\u6709\u663e\u8457\u63d0\u5347\uff0c\u4f8b\u5982\uff0cGPT-4o-mini\u7684\u51c6\u786e\u7387\u63d0\u5347\u8d85\u8fc7\u56db\u500d\u3002<\/p>\n\n\n\n

\u5c3d\u7ba1\u6d4f\u89c8\u80fd\u529b\u5e26\u6765\u4e86\u663e\u8457\u6539\u8fdb\uff0c\u4f46\u5373\u4fbf\u662f\u6700\u5148\u8fdb\u7684\u641c\u7d22\u589e\u5f3a\u578b\u6a21\u578b\uff0c\u5982GPT-4o-search-preview\uff0c\u5176\u51c6\u786e\u7387\u4ec5\u4e3a18.83%<\/strong>\u3002<\/p>\n\n\n\n

\u4e8c\u3001\u65b9\u6cd5<\/p>\n\n\n\n

OpenAI\u7684Deep Research\u3001Grok\u7684DeepSearch\u3001Gemini\u7684Deep Research\u4ee5\u53ca\u6708\u4e4b\u6697\u9762\u7684Kimi-Researcher\u7b49\uff0c\u4ee5\u201c\u6df1\u5ea6\u641c\u7d22\u201d\u529f\u80fd\u4e3a\u6838\u5fc3\uff0c\u4e3a\u653b\u514b\u9ad8\u96be\u5ea6\u4fe1\u606f\u68c0\u7d22\u4efb\u52a1\u63d0\u4f9b\u4e86\u65b0\u7684\u8303\u5f0f\u3002<\/p>\n\n\n\n

\u7136\u800c\uff0c\u5b66\u672f\u754c\u4e0e\u4e1a\u754c\u76ee\u524d\u5c1a\u672a\u5efa\u7acb\u8d77\u4e00\u5957\u516c\u8ba4\u7684\u8bc4\u4f30\u4f53\u7cfb\u4e0e\u6807\u51c6\u6570\u636e\u96c6\uff0c\u7528\u4ee5\u7cfb\u7edf\u6027\u5730\u68c0\u9a8c\u8fd9\u4e9b\u65b0\u5174\u6a21\u578b\u5728\u771f\u5b9e\u5b66\u672f\u7814\u7a76\u573a\u666f<\/strong>\u4e0b\u7684\u5b9e\u9645\u6548\u80fd\u3002<\/p>\n\n\n\n

\u5317\u4eac\u5927\u5b66DS-Lab<\/strong>\u53d1\u5e03ScholarSearch\uff0c<\/strong>\u65e8\u5728\u5bf9LLMs\u7684\u68c0\u7d22\u3001\u4fe1\u606f\u6574\u5408\u53ca\u63a8\u7406\u80fd\u529b\u8fdb\u884c\u7efc\u5408\u6027\u3001\u6781\u9650\u6027\u8003\u9a8c\u3002<\/p>\n\n\n\n

\u7814\u7a76\u56e2\u961f\u62db\u52df\u4e86\u6765\u81ea\u5317\u4eac\u5927\u5b66\u5404\u4e2a\u5b66\u9662\u7684\u672c\u79d1\u548c\u7814\u7a76\u751f\u5fd7\u613f\u8005\uff0c\u5e76\u4e3a\u4ed6\u4eec\u63d0\u4f9b\u4e86\u96c6\u4e2d\u57f9\u8bad\u3002\u5fd7\u613f\u8005\u4ece\u516c\u5f00\u53ef\u8bbf\u95ee\u7684\u5728\u7ebf\u51fa\u7248\u7269\u548c\u7f51\u7ad9\u4e2d\u9009\u62e9\u6750\u6599\uff0c\u4ee5\u5236\u5b9a\u9700\u8981\u7f51\u7edc\u641c\u7d22\u89e3\u7b54\u7684\u5b66\u672f\u95ee\u9898\u3002<\/p>\n\n\n\n

\u4e3a\u786e\u4fdd\u95ee\u9898\u80fd\u771f\u6b63\u8003\u9a8c\u6a21\u578b\u7684\u6df1\u5ea6\u7814\u7a76\u80fd\u529b\uff0c\u6240\u6709\u521d\u6b65\u6784\u5efa\u7684\u95ee\u9898\u5fc5\u987b\u901a\u8fc7\u4ee5\u4e0b\u53cc\u91cd\u8d1f\u5411\u7b5b\u9009\u6807\u51c6\u7684\u9a8c\u8bc1\uff1a<\/p>\n\n\n\n

1.\u4e0d\u80fd\u901a\u8fc7Grok 3\u7684Thinking\u6a21\u5f0f\u83b7\u5f97\u6b63\u786e\u7b54\u6848\uff0c\u786e\u4fdd\u95ee\u9898\u9700\u8981\u6df1\u5165\u5e7f\u6cdb\u7684\u4fe1\u606f\u68c0\u7d22\u80fd\u529b\u3002
2.Grok 3\u7684DeepSearch\u6a21\u5f0f\u6216Gemini 2.5 Pro\u7684Deep Research\u529f\u80fd\u81f3\u5c11\u6709\u4e00\u4e2a\u672a\u80fd\u63d0\u4f9b\u6b63\u786e\u7b54\u6848\uff0c\u786e\u4fdd\u95ee\u9898\u7684\u9ad8\u96be\u5ea6\u3002<\/p>\n\n\n\n

\u6210\u529f\u6ee1\u8db3\u4e0a\u8ff0\u6807\u51c6\u7684\u95ee\u9898\u968f\u540e\u63d0\u4ea4\u7ed9\u4e13\u95e8\u7684\u5ba1\u6838\u56e2\u961f\u8fdb\u884c\u6570\u636e\u5ba1\u6838\uff0c\u4ee5\u786e\u4fdd\u4ee5\u4e0b\u51e0\u70b9\uff1a<\/p>\n\n\n\n

\u7b54\u6848\u552f\u4e00\u6027<\/strong>\uff1a\u6bcf\u4e2a\u95ee\u9898\u5bf9\u5e94\u552f\u4e00\u7684\u3001\u660e\u786e\u65e0\u8bef\u7684\u7b54\u6848\u3002
\u6765\u6e90\u53ef\u8bbf\u95ee\u6027<\/strong>\uff1a\u56de\u7b54\u95ee\u9898\u6240\u9700\u7684\u53c2\u8003\u6765\u6e90\u53ef\u901a\u8fc7\u4e92\u8054\u7f51\u516c\u5f00\u83b7\u53d6\u3002
\u5b66\u672f\u6b63\u786e\u6027<\/strong>\uff1a\u95ee\u9898\u7684\u5b66\u672f\u4ef7\u503c\u548c\u7b54\u6848\u7684\u6b63\u786e\u6027\u6839\u636e\u63d0\u4f9b\u7684\u6765\u6e90\u8fdb\u884c\u9a8c\u8bc1\u3002<\/p>\n\n\n\n

\u4efb\u4f55\u672a\u80fd\u8fbe\u6807\u7684\u95ee\u9898\u90fd\u5c06\u88ab\u9000\u56de\u8fdb\u884c\u8fed\u4ee3\u4fee\u8ba2\uff0c\u76f4\u81f3\u5408\u683c\u3002<\/p>\n\n\n\n

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

\u4e09\u3001\u6570\u636e\u96c6\u6838\u5fc3\u7279\u70b9<\/p>\n\n\n\n

\u9ad8\u5ea6\u7684\u771f\u5b9e\u6027\u4e0e\u5e94\u7528\u4ef7\u503c<\/strong>\uff1a\u6570\u636e\u96c6\u4e2d\u7684\u6240\u6709\u95ee\u9898\u5747\u6e90\u4e8e\u771f\u5b9e\u7684\u5b66\u672f\u7814\u7a76\u4e0e\u5b66\u4e60\u60c5\u5883\u3002\u5176\u8bbe\u8ba1\u65e8\u5728\u5fe0\u5b9e\u53cd\u6620\u7814\u7a76\u8005\u9762\u4e34\u7684\u5b9e\u9645\u4fe1\u606f\u6311\u6218\uff0c\u4fdd\u8bc1\u4e86\u8bc4\u4f30\u7ed3\u679c\u80fd\u591f\u771f\u5b9e\u5730\u53cd\u6620\u6a21\u578b\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\u7684\u6548\u80fd\u3002<\/p>\n\n\n\n

\u5353\u8d8a\u7684\u6311\u6218\u6027\u4e0e\u6df1\u5ea6<\/strong>\uff1a<\/strong>ScholarSearch\u7684\u96be\u5ea6\u7ecf\u8fc7\u4e25\u683c\u5ba1\u67e5\uff0c\u786e\u4fdd\u5373\u4fbf\u662f\u9876\u5c16\u7684\u6a21\u578b\uff08\u5982 Grok DeepSearch \u6216 Gemini Deep Research\uff09\u4e5f\u96be\u4ee5\u4e00\u6b21\u6027\u7ed9\u51fa\u6b63\u786e\u7b54\u6848\u3002\u5927\u591a\u6570\u95ee\u9898\u9700\u8981\u8fdb\u884c\u591a\u6b21\u6df1\u5ea6\u641c\u7d22\u624d\u80fd\u5f97\u51fa\u7b54\u6848\uff0c\u5145\u5206\u8003\u9a8c\u6a21\u578b\u7684\u590d\u6742\u63a8\u7406\u548c\u4fe1\u606f\u6574\u5408\u80fd\u529b\u3002<\/p>\n\n\n\n

\u5e7f\u6cdb\u7684\u5b66\u79d1\u8986\u76d6<\/strong>\uff1a\u4e3a\u786e\u4fdd\u8bc4\u4f30\u7684\u5168\u9762\u6027\u4e0e\u4ee3\u8868\u6027\uff0cScholarSearch\u56f4\u7ed5\u79d1\u5b66\u4e0e\u5de5\u7a0b\u9886\u57df\uff08Science & Engineering\uff09\u548c\u793e\u4f1a\u79d1\u5b66\u4e0e\u4eba\u6587\u5b66\u79d1\u9886\u57df\uff08Social Sciences & Humanities\uff09\u4e24\u5927\u95e8\u7c7b\u8fdb\u884c\u6784\u5efa\uff0c\u5171\u6db5\u76d6\u4e8615\u4e2a\u4e0d\u540c\u7684\u7ec6\u5206\u5b66\u79d1\u3002<\/p>\n\n\n\n

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

\u56db\u3001\u7ed3\u679c<\/p>\n\n\n\n

\u7814\u7a76\u56e2\u961f\u4f7f\u7528ScholarSearch\u5bf9\u5177\u5907\u8054\u7f51\u641c\u7d22\u80fd\u529b\u7684\u4ee3\u8868\u6027\u6a21\u578b\u53ca\u7eaf\u63a8\u7406\u6a21\u578b\u8fdb\u884c\u4e86\u8bc4\u4f30\uff0c\u7ed3\u679c\u5982\u56fe\u6240\u793a\u3002\u8bc4\u4f30\u7ed3\u679c\u660e\u786e\u63ed\u793a\uff0c\u73b0\u6709\u6a21\u578b\u7684\u6574\u4f53\u8868\u73b0\u6b20\u4f73\uff0c\u5176\u5b66\u672f\u641c\u7d22\u80fd\u529b\u4e9f\u5f85\u63d0\u5347\u3002<\/p>\n\n\n\n

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

\u4e94\u3001\u603b\u7ed3<\/p>\n\n\n\n

\u4ec5\u51ed\u63a8\u7406\u65e0\u6cd5\u89e3\u51b3\u5b66\u672f\u7814\u7a76\u95ee\u9898<\/strong>\uff1a \u5b9e\u9a8c\u660e\u786e\u6307\u51fa\uff0cScholarSearch\u6570\u636e\u96c6\u4e2d\u7684\u95ee\u9898\u65e0\u6cd5\u4ec5\u901a\u8fc7\u6a21\u578b\u7684\u9884\u8bad\u7ec3\u77e5\u8bc6\u548c\u63a8\u7406\u80fd\u529b\u89e3\u51b3\u3002\u9876\u5c16\u7684\u7eaf\u63a8\u7406\u6a21\u578b\uff0c\u5982GPT-4.1\u3001DeepSeek-R1\uff0c\u5728\u5904\u7406\u8fd9\u4e9b\u95ee\u9898\u65f6\u51c6\u786e\u7387\u6781\u4f4e\uff0c\u666e\u904d\u4f4e\u4e8e9%\u3002\u8fd9\u8868\u660e\u5b66\u672f\u67e5\u8be2\u5177\u6709\u9ad8\u5ea6\u590d\u6742\u6027\uff0c\u8fd9\u8d85\u51fa\u4e86\u9759\u6001\u3001\u5185\u5d4c\u77e5\u8bc6\u5e93\u7684\u80fd\u529b\u8303\u7574\u3002<\/p>\n\n\n\n

\u6d4f\u89c8\u80fd\u529b\u663e\u8457\u63d0\u5347\u6a21\u578b\u6027\u80fd<\/strong>\uff1a \u8d4b\u4e88\u6a21\u578b\u8bbf\u95ee\u4e92\u8054\u7f51\u7684\u6d4f\u89c8\u80fd\u529b\u53ef\u4ee5\u63d0\u9ad8\u5176\u51c6\u786e\u6027\u3002\u5177\u5907\u641c\u7d22\u529f\u80fd\u7684\u6a21\u578b\uff0c\u76f8\u8f83\u4e8e\u5176\u65e0\u641c\u7d22\u80fd\u529b\u7684\u7248\u672c\uff0c\u51c6\u786e\u7387\u6709\u663e\u8457\u63d0\u5347\uff0c\u4f8b\u5982\uff0cGPT-4o-mini\u7684\u51c6\u786e\u7387\u63d0\u5347\u8d85\u8fc7\u56db\u500d\u3002\u6b64\u5916\uff0c\u641c\u7d22\u80fd\u529b\u4e5f\u5e73\u8861\u4e86\u6a21\u578b\u5728\u4e0d\u540c\u5b66\u79d1\u9886\u57df\u7684\u8868\u73b0\uff0c\u5728\u79d1\u5b66\u4e0e\u5de5\u7a0b\u9886\u57df\u548c\u793e\u4f1a\u79d1\u5b66\u4e0e\u4eba\u6587\u5b66\u79d1\u9886\u57df\u8fbe\u5230\u4e86\u76f8\u5f53\u7684\u6c34\u5e73\u3002\u8fd9\u4e00\u7ed3\u679c\u8bc1\u5b9e\uff0c\u5bf9\u4e8e\u89e3\u51b3\u590d\u6742\u7684\u5b66\u672f\u95ee\u9898\uff0c\u8fdb\u884c\u5b9e\u65f6\u4fe1\u606f\u68c0\u7d22\u3001\u8bbf\u95ee\u6570\u636e\u5e76\u8fdb\u884c\u4ea4\u53c9\u5f15\u7528\u7684\u80fd\u529b\u81f3\u5173\u91cd\u8981\u3002<\/p>\n\n\n\n

\u5f53\u524d\u641c\u7d22\u6a21\u578b\u4ecd\u4e0d\u8db3\u4ee5\u5e94\u5bf9\u6df1\u5ea6\u5b66\u672f\u63a2\u7a76<\/strong>\uff1a \u5c3d\u7ba1\u6d4f\u89c8\u80fd\u529b\u5e26\u6765\u4e86\u663e\u8457\u6539\u8fdb\uff0c\u4f46\u5373\u4fbf\u662f\u6700\u5148\u8fdb\u7684\u641c\u7d22\u589e\u5f3a\u578b\u6a21\u578b\uff0c\u5982GPT-4o-search-preview\uff0c\u51c6\u786e\u7387\u4ec5\u4e3a18.83%\u3002\u5728\u89e3\u51b3\u590d\u6742\u5b66\u672f\u95ee\u9898\u65b9\u9762\u4ecd\u8868\u73b0\u4e0d\u5145\u5206\u3002\u8fd9\u63ed\u793a\u4e86\u5f53\u524d\u6a21\u578b\u5728\u8fdb\u884c\u6df1\u5ea6\u7814\u7a76\u3001\u6574\u5408\u4e13\u4e1a\u77e5\u8bc6\u4ee5\u53ca\u6267\u884c\u590d\u6742\u7684\u591a\u6e90\u63a8\u7406\u65f6\u5b58\u5728\u7684\u5dee\u8ddd\uff0c\u4e5f\u63ed\u793a\u4e86Deep Research\u6a21\u578b\u7684\u7814\u53d1\u9700\u6c42\u3002<\/p>\n\n\n\n

ScholarSearch\u4f5c\u4e3a\u4e00\u4e2a\u5728\u6df1\u5ea6\u641c\u7d22\u9886\u57df\u7684\u5b66\u672f\u57fa\u51c6\u6d4b\u8bd5\u96c6\uff0c\u4e0d\u4ec5\u8861\u91cf\u4e86\u6a21\u578b\u7684\u5f53\u524d\u80fd\u529b\uff0c\u66f4\u63ed\u793a\u4e86\u73b0\u6709\u6280\u672f\u4e0e\u771f\u5b9e\u5b66\u672f\u5de5\u4f5c\u6d41\u4e4b\u95f4\u7684\u6838\u5fc3\u5dee\u8ddd\uff0c\u4e3a\u672a\u6765\u7684\u5927\u8bed\u8a00\u6a21\u578b\u638c\u63e1\u590d\u6742\u7efc\u5408\u7684\u8bed\u5883\u7406\u89e3\u3001\u6d77\u91cf\u8d44\u6599\u6765\u6e90\u7684\u6279\u5224\u51c6\u786e\u6027\u9a8c\u8bc1\uff0c\u63d0\u4f9b\u4e86\u6709\u6311\u6218\u7684\u53c2\u8003\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

\u5355\u4f4d\uff1aArxiv \u4f5c\u8005\uff1aJunting Zhou*, Wang Li*, Yiyan Liao*,Nengyu […]<\/p>\n","protected":false},"author":9,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-3688","post","type-post","status-publish","format-standard","hentry","category-events"],"_links":{"self":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3688","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\/9"}],"replies":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/comments?post=3688"}],"version-history":[{"count":1,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3688\/revisions"}],"predecessor-version":[{"id":3695,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/3688\/revisions\/3695"}],"wp:attachment":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/media?parent=3688"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/categories?post=3688"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/tags?post=3688"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}