{"id":10466,"date":"2026-07-26T22:32:47","date_gmt":"2026-07-26T14:32:47","guid":{"rendered":"https:\/\/infernews.com\/blog\/a-bilingual-chinese-english-benchmark-for-document-grounded-complex-financial-qa\/"},"modified":"2026-07-26T22:33:51","modified_gmt":"2026-07-26T14:33:51","slug":"a-bilingual-chinese-english-benchmark-for-document-grounded-complex-financial-qa","status":"publish","type":"post","link":"https:\/\/infernews.com\/blog\/a-bilingual-chinese-english-benchmark-for-document-grounded-complex-financial-qa\/","title":{"rendered":"FinanceComplexQA \u9ede\u8a55\uff1a\u91d1\u878d\u9577\u6587\u4ef6\u554f\u7b54\u57fa\u6e96"},"content":{"rendered":"\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" src=\"https:\/\/infernews.com\/blog\/wp-content\/uploads\/2026\/07\/pasted-64e245cbc960.jpg\" alt=\"Finance-ComplexQA at a Glance\"><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">\u91d1\u878d\u554f\u7b54\u6700\u5bb9\u6613\u5931\u771f\u7684\u4f4d\u7f6e\uff0c\u4e0d\u662f\u6a21\u578b\u8b58\u5514\u8b58\u8853\u8a9e\uff0c\u800c\u662f\u5b83\u6703\u5426\u771f\u6b63\u5728\u6574\u4efd\u53c3\u8003\u6587\u4ef6\u5165\u9762\u63a8\u7406\u3001\u6bd4\u5c0d\u540c\u8a08\u6578\u3002FinanceComplexQA\u5c6c\u65bc<strong>\u6578\u64da\u96c6\uff0fBenchmark<\/strong>\uff0c\u7126\u9ede\u4e0d\u662f\u80cc\u7b54\u6848\uff0c\u800c\u662f\u6aa2\u9a57 LLMs \u548c agents \u80fd\u5426\u6839\u64da\u5b8c\u6574 reference documents \u56de\u7b54\u8907\u96dc\u91d1\u878d\u554f\u984c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u5b83\u4fee\u6b63\u4e86\u53ea\u9760 parametric knowledge \u6216\u62bd\u53d6\u55ae\u4e00\u6bb5\u843d\u7684\u8a55\u6e2c\u7bc4\u5f0f\u3002\u4f5c\u8005\u628a\u91cd\u9ede\u653e\u5728 document-grounded complex financial QA\uff0c\u8981\u6c42\u7b54\u6848\u540c\u554f\u984c\u53ca\u539f\u59cb\u6587\u4ef6\u4e00\u81f4\uff0c\u4e26\u6db5\u84cb multi-hop reasoning\u3001numerical calculation\u3001comparison\u3001implicit inference\u3001planning\u3001summarization \u540c evidence-grounded verification\uff0c\u5c0d RAG\u3001Agentic workflow \u540c\u9577\u6587\u672c\u95b1\u8b80\u80fd\u529b\u90fd\u6709\u53c3\u8003\u50f9\u503c\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u8cc7\u6599\u7d50\u69cb\u672c\u8eab\u4ea6\u6709\u53d6\u6368\u3002FinComplexQA-Pro \u6536\u9304 2,026 \u7d44\u7368\u7acb QA\uff0c\u6309\u8a9e\u8a00\u3001\u91d1\u878d\u5834\u666f\u8207\u4efb\u52d9\u5206\u985e\u7d44\u7e54\uff1b\u540c\u4e00\u984c\u6703\u4ee5 scene_categories \u8207 task_categories \u5169\u7a2e\u8996\u89d2\u51fa\u73fe\uff0c\u6240\u4ee5\u7e3d\u8a18\u9304\u8996\u5716\u6709 4,052 \u7b46\u3002\u53e6\u6709 overall \u63d0\u4f9b agent_answer\u3001agent_thinking \u53ca LLM-as-a-judge \u5206\u6578\uff0c\u4f46\u9019\u4e9b\u5206\u6578\u53ea\u9069\u5408\u505a\u8a3a\u65b7\u8a0a\u865f\uff0c\u4e0d\u80fd\u7576 ground truth\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u652f\u63f4\u4e2d\u6587\u8207\u82f1\u6587\uff0c\u4f46\u5169\u500b\u5b50\u96c6\u8986\u84cb\u7684\u6587\u4ef6\u9818\u57df\u4e0d\u540c\uff0cschema \u4ea6\u4e0d\u5b8c\u5168\u4e00\u81f4<\/li>\n<li>\u8f03\u9069\u5408\u9010\u500b\u5b50\u76ee\u9304\u8b80\u53d6 JSONL\uff0c\u800c\u4e0d\u662f\u4e00\u958b\u59cb\u5408\u4f75\u5168\u90e8\u8cc7\u6599<\/li>\n<li>\u53ef\u7528 exact match\u3001\u6578\u503c\u5bb9\u5dee\u3001F1\u3001semantic similarity \u7b49\u65b9\u6cd5\u6bd4\u5c0d\u8f38\u51fa<\/li>\n<li>\u9644\u6709 Reference_documents\uff0c\u65b9\u4fbf\u8ffd\u67e5 PDF \u8207 LaTeX \u539f\u6587\u8b49\u64da<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\">\u90e8\u7f72\u548c\u6e2c\u8a66\u7684\u7406\u89e3\u65b9\u5f0f\u76f8\u7576\u76f4\u63a5\uff1a\u8cc7\u6599\u4e3b\u8981\u5728 Hugging Face \u767c\u4f48\uff0c\u7814\u7a76\u5718\u968a\u53ef\u5148\u6311\u55ae\u4e00\u8a9e\u8a00\u3001\u55ae\u4e00 task category \u8f09\u5165\uff0c\u518d\u628a\u6a21\u578b\u8f38\u51fa\u5c0d\u7167 gold answer \u6216\u6587\u4ef6\u8b49\u64da\u505a\u8a55\u4f30\u3002\u5b83\u8f03\u53d7\u60e0\u65bc\u505a\u91d1\u878d RAG\u3001\u9577\u6587\u4ef6 QA\u3001Agent \u8a55\u6e2c\u6216\u96d9\u8a9e\u7814\u7a76\u7684\u5718\u968a\uff1b\u8981\u7559\u610f\u7684\u662f\u91d1\u878d\u4e8b\u5be6\u5177\u6642\u6548\u6027\uff0c\u800c\u4e14\u9805\u76ee\u5df2\u660e\u78ba\u6a19\u793a\u50c5\u4f9b\u7814\u7a76\u8207\u8a55\u4f30\uff0c\u4e0d\u61c9\u5ef6\u4f38\u6210\u6295\u8cc7\u3001\u6703\u8a08\u3001\u6cd5\u5f8b\u6216\u8ca1\u52d9\u5efa\u8b70\u3002<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><a href=\"https:\/\/huggingface.co\/datasets\/Multilingual-Multimodal-NLP\/FinanceComplexQA\" rel=\"noopener noreferrer\"><strong>\u9805\u76ee\u4e3b\u9801<\/strong><\/a> \u00b7 <a href=\"https:\/\/github.com\/buaacxf\/FinanceComplexQA\" rel=\"noopener noreferrer\"><strong>GitHub<\/strong><\/a> \u00b7 <a href=\"https:\/\/arxiv.org\/pdf\/2607.19238\" rel=\"noopener noreferrer\"><strong>Paper<\/strong><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u60f3\u6e2c\u8a66\u6a21\u578b\u662f\u5426\u771f\u4fc2\u8b80\u5f97\u61c2\u8ca1\u52d9\u6587\u4ef6\uff0cFinanceComplexQA\u6bd4\u4e00\u822c\u62bd\u53e5\u5f0f\u554f\u7b54\u66f4\u8cbc\u8fd1\u96e3\u984c\u3002\u5b83\u628a\u4e2d\u82f1\u96d9\u8a9e\u3001\u63a8\u7406\u3001\u591a\u6b65\u8a08\u7b97\u8207\u6587\u4ef6\u4f9d\u64da\u7d81\u5728\u540c\u4e00\u500b\u57fa\u6e96\u5167\u3002<\/p>\n","protected":false},"author":8,"featured_media":10465,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ai_generated_summary":"","footnotes":""},"categories":[133,173,147,116,109,119,189,199],"tags":[],"class_list":["post-10466","post","type-post","status-publish","format-standard","hentry","category-133","category-173","category-deepseek","category-agentic","category-rag","category-119","category-189","category-dataset-"],"_links":{"self":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10466","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/comments?post=10466"}],"version-history":[{"count":1,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10466\/revisions"}],"predecessor-version":[{"id":10468,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/posts\/10466\/revisions\/10468"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media\/10465"}],"wp:attachment":[{"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/media?parent=10466"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/categories?post=10466"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/infernews.com\/blog\/wp-json\/wp\/v2\/tags?post=10466"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}