Files
rag/tests/eval_dataset_v2.json
lacerate551 8c92657490 fix(bm25): 修复 BM25 索引覆盖 bug + 综合评测集 v2 + 检索增强
核心修复:
- knowledge/base.py: BM25Index.add_documents 从覆盖改为追加+去重,
  修复只有最后上传文件的 chunks 保留在 BM25 中的严重 bug
  (影响: 2.docx/3.docx/PDF 的 755 个 chunk 在 BM25 中完全缺失)

检索增强 (延续上次会话):
- core/engine.py: section cluster boost + lexical match exemption
- api/chat_routes.py: lexical/cluster rescue 层 + SSE 事件
- core/mmr.py: MMR 去重改进

评测体系:
- tests/eval_dataset_v2.json: 62 题综合评测集 (9 种题型×4 文档)
- scripts/eval_e2e.py: 推理模型 LLM 评分兼容 + 新数据集格式支持
- scripts/validate_eval_dataset.py: 数据集验证工具

其他:
- parsers/mineru_parser.py: 解析器改进
2026-06-17 17:31:43 +08:00

615 lines
30 KiB
JSON
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
{
"description": "RAG 系统综合评测数据集 v2 — 覆盖货源投放、吸烟环境、零售终端、三峡公报四个知识库文档",
"version": "2.0",
"created_at": "2026-06-17",
"source_documents": [
"1.docx",
"2.docx",
"3.docx",
"三峡公报_1-15页.pdf"
],
"query_categories": [
"simple_fact",
"enumeration",
"definition",
"comparison",
"reasoning",
"table_data",
"cross_doc",
"negative",
"paraphrase"
],
"queries": [
{
"id": "q001",
"query": "货源投放有哪些方式?",
"query_type": "enumeration",
"relevant_docs": ["1.docx"],
"reference_answer": "货源投放方式主要有六种1.按档位投放2.按档位+标签扩展投放3.按价位段自选投放4.选点投放5.批零网配6.事件用烟投放。",
"expected_keywords": ["按档位投放", "标签扩展", "价位段自选", "选点投放", "批零网配", "事件用烟", "六种"],
"difficulty": "easy"
},
{
"id": "q002",
"query": "市场状态分为哪几种?分别是什么?",
"query_type": "enumeration",
"relevant_docs": ["1.docx"],
"reference_answer": "市场状态分为五种:俏、紧、平、松、软。",
"expected_keywords": ["俏", "紧", "平", "松", "软", "五种"],
"difficulty": "easy"
},
{
"id": "q003",
"query": "什么是紧俏品规?",
"query_type": "definition",
"relevant_docs": ["1.docx"],
"reference_answer": "紧俏品规是指当地消费需求持续旺盛货源缺口较大零售价格坚挺在较长时间内不能满足市场需求的品规。紧俏品规数不得超过地市级公司所经营品规数的20%。",
"expected_keywords": ["消费需求", "旺盛", "货源缺口", "零售价格坚挺", "20%"],
"difficulty": "medium"
},
{
"id": "q004",
"query": "紧俏品规数量有什么限制?",
"query_type": "simple_fact",
"relevant_docs": ["1.docx"],
"reference_answer": "紧俏品规数不得超过地市级公司所经营品规数的20%。",
"expected_keywords": ["20%", "不得超过", "品规数"],
"difficulty": "easy"
},
{
"id": "q005",
"query": "均衡满足品规有什么限制?",
"query_type": "simple_fact",
"relevant_docs": ["1.docx"],
"reference_answer": "均衡满足品规数不得超过地市级公司所经营顺销品规数的40%(顺销品规包括均衡满足品规和完全满足品规)。",
"expected_keywords": ["40%", "不得超过", "顺销品规", "均衡满足"],
"difficulty": "easy"
},
{
"id": "q006",
"query": "单客户单次单品规订货上限是多少?",
"query_type": "simple_fact",
"relevant_docs": ["1.docx"],
"reference_answer": "单客户单次单品规订货上限不超过50条。",
"expected_keywords": ["50条", "单客户", "单品规", "不超过"],
"difficulty": "easy"
},
{
"id": "q007",
"query": "月度单客户订货总量上限是怎么规定的?",
"query_type": "simple_fact",
"relevant_docs": ["1.docx"],
"reference_answer": "月度单客户订货总量上限为上一年度平均月度进货量的15倍。",
"expected_keywords": ["15倍", "平均月度进货量", "上一年度"],
"difficulty": "easy"
},
{
"id": "q008",
"query": "主导品规的周供应占比要求是多少?",
"query_type": "simple_fact",
"relevant_docs": ["1.docx"],
"reference_answer": "主导品规周供应量应达到本价位段周供应量的60%以上。",
"expected_keywords": ["60%", "价位段", "周供应量", "主导品规"],
"difficulty": "medium"
},
{
"id": "q009",
"query": "主导品规和护卫品规的供应占比合计要求是多少?",
"query_type": "simple_fact",
"relevant_docs": ["1.docx"],
"reference_answer": "主导品规和护卫品规的周供应量合计应达到本价位段周供应量的85%以上。",
"expected_keywords": ["85%", "主导品规", "护卫品规", "合计"],
"difficulty": "medium"
},
{
"id": "q010",
"query": "货源投放的总体要求是什么?",
"query_type": "enumeration",
"relevant_docs": ["1.docx"],
"reference_answer": "货源投放是烟草营销的核心业务,总体要求包括六个坚持:坚持市场导向、供需匹配;坚持总量控制、稍紧平衡;坚持增速合理、贵在持续;坚持公平公正、严格规范;坚持状态优先、科学投放;坚持区域协同、高效运作。",
"expected_keywords": ["市场导向", "总量控制", "稍紧平衡", "公平公正", "状态优先", "区域协同", "六个"],
"difficulty": "medium"
},
{
"id": "q011",
"query": "新品投放有什么要求和限制?",
"query_type": "reasoning",
"relevant_docs": ["1.docx"],
"reference_answer": "新品指二类及以上销售时间不超过24个月的品规。新品可采用选点投放方式要求认真分析新品定位坚持科学民主决策制订具体选点标准不得人为指定选点客户名单。",
"expected_keywords": ["24个月", "选点投放", "科学民主", "不得人为指定"],
"difficulty": "medium"
},
{
"id": "q012",
"query": "紧俏品规在投放时有什么特殊要求?",
"query_type": "reasoning",
"relevant_docs": ["1.docx"],
"reference_answer": "紧俏品规投放要更加注重公平投放、普惠性投放,适度扩大供货面,且不得按档位扩展投放。",
"expected_keywords": ["公平投放", "普惠", "扩大供货面", "不得", "档位扩展"],
"difficulty": "hard"
},
{
"id": "q013",
"query": "按档位投放和选点投放有什么区别?",
"query_type": "comparison",
"relevant_docs": ["1.docx"],
"reference_answer": "按档位投放是由系统根据客户档级自动分配投放量,面向所有在档客户。选点投放是针对特定客户群体进行的定向投放,需要制定选点标准,经过集体决策确定投放对象。",
"expected_keywords": ["档级", "自动分配", "选点", "定向", "集体决策"],
"difficulty": "medium"
},
{
"id": "q014",
"query": "主导品规和护卫品规有什么区别?",
"query_type": "comparison",
"relevant_docs": ["1.docx"],
"reference_answer": "主导品规一般有1-4个属于深受本地消费者喜爱、零售客户欢迎的品规。护卫品规一般有2-5个属于本地有相当数量消费者喜爱的品规。主要区别在于规模和市场地位不同主导品规规模更大、地位更重要。",
"expected_keywords": ["1-4个", "2-5个", "消费者喜爱", "规模", "市场地位"],
"difficulty": "medium"
},
{
"id": "q015",
"query": "如果某个品规供货面高、订货面低,应该怎么调整投放策略?",
"query_type": "reasoning",
"relevant_docs": ["1.docx"],
"reference_answer": "供货面高但订货面低说明投放范围过大但客户需求不足。应缩小供货面,减少投放客户范围,将货源集中到有真实需求的客户。",
"expected_keywords": ["供货面", "订货面", "调整", "策略"],
"difficulty": "hard"
},
{
"id": "q016",
"query": "批零网配客户配货增加量有什么限制?",
"query_type": "table_data",
"relevant_docs": ["1.docx"],
"reference_answer": "批零网配客户配货增加量不得超过同档级客户初始合理定量的50%。",
"expected_keywords": ["50%", "同档级", "初始合理定量", "不得超过"],
"difficulty": "medium"
},
{
"id": "q017",
"query": "市场状态评价指标体系中各指标的权重是怎样的?",
"query_type": "table_data",
"relevant_docs": ["1.docx"],
"reference_answer": "市场状态评价采用双层双级指标体系,不同品类(紧俏品规、均衡满足品规、宽松品规)的评价指标权重不同,包括零售价格指数、市场流通价格指数、终端动销率、终端库存、社会存销比、订货面、订足面等指标。",
"expected_keywords": ["双层双级", "权重", "零售价格指数", "动销率", "存销比", "订货面"],
"difficulty": "hard"
},
{
"id": "q018",
"query": "文明吸烟环境的SUCCESS功能框架包含哪些功能",
"query_type": "enumeration",
"relevant_docs": ["2.docx"],
"reference_answer": "SUCCESS功能框架包含7个功能S-Smoking吸烟、U-Utility实用、C-Convenience便利、C-Comfortable舒适、E-Experience体验、S-Safe安全、S-Survey调研。",
"expected_keywords": ["Smoking", "Utility", "Convenience", "Comfortable", "Experience", "Safe", "Survey", "七个"],
"difficulty": "medium"
},
{
"id": "q019",
"query": "吸烟设施有哪些类型?",
"query_type": "enumeration",
"relevant_docs": ["2.docx"],
"reference_answer": "吸烟设施分为三种类型:吸烟室、吸烟区、吸烟点。",
"expected_keywords": ["吸烟室", "吸烟区", "吸烟点", "三种"],
"difficulty": "easy"
},
{
"id": "q020",
"query": "吸烟环境分为哪几个区域类别?各覆盖什么场所?",
"query_type": "enumeration",
"relevant_docs": ["2.docx"],
"reference_answer": "分为三大类九个子类A类公共服务区域包括A1交通枢纽、A2政务服务、A3旅游景区、A4医疗机构B类商业区域包括B1商场、B2酒店、B3餐饮娱乐C类办公/工业区域包括C1办公楼、C2工业园区。",
"expected_keywords": ["A类", "B类", "C类", "交通枢纽", "政务", "景区", "商场", "酒店", "办公楼"],
"difficulty": "hard"
},
{
"id": "q021",
"query": "吸烟场所的建设标准是哪一年发布的?共有多少条?",
"query_type": "simple_fact",
"relevant_docs": ["2.docx"],
"reference_answer": "文明吸烟环境建设标准试行共7章12条管理标准试行共8章20条维护标准试行共6章9条。三份标准均为试行版本。",
"expected_keywords": ["试行", "建设标准", "管理标准", "维护标准", "12条", "20条", "9条"],
"difficulty": "medium"
},
{
"id": "q022",
"query": "吸烟室、吸烟区和吸烟点有什么区别?",
"query_type": "comparison",
"relevant_docs": ["2.docx"],
"reference_answer": "吸烟室是独立的封闭空间,有面积和通风要求;吸烟区是在开放或半开放空间中划定的区域;吸烟点是最小化的设施,通常只配备烟灰缸和标识牌。三者在面积、设施配备、适用场所上有所不同。",
"expected_keywords": ["封闭空间", "开放", "划定", "面积", "设施配备", "场所"],
"difficulty": "medium"
},
{
"id": "q023",
"query": "A类区域和B类区域的吸烟设施配置有什么不同要求",
"query_type": "comparison",
"relevant_docs": ["2.docx"],
"reference_answer": "A类区域公共服务区域和B类区域商业区域对吸烟设施的类型选择、面积标准、设备配置等有不同的要求。A类区域通常要求更高标准的设施配置。",
"expected_keywords": ["A类", "B类", "配置", "面积", "标准", "不同"],
"difficulty": "hard"
},
{
"id": "q024",
"query": "吸烟设施的日常维护和定期维护有什么区别?",
"query_type": "comparison",
"relevant_docs": ["2.docx"],
"reference_answer": "日常维护是指每日或高频次的保洁、设备检查等基础工作。定期维护是指按固定周期进行的深度清洁、设备更换、设施检修等工作。两者的频率、深度和负责人员不同。",
"expected_keywords": ["日常维护", "定期维护", "频率", "保洁", "设备更换", "深度清洁"],
"difficulty": "medium"
},
{
"id": "q025",
"query": "吸烟设施的编码规则是怎样的?",
"query_type": "definition",
"relevant_docs": ["2.docx"],
"reference_answer": "吸烟设施编码格式为:所在区(县/市)+街道(镇)+道路名称+场所编码吸烟室为AXXX、吸烟区亭为BXXX、吸烟点为CXXX+具体设备编码XXX。",
"expected_keywords": ["编码", "区县", "街道", "AXXX", "BXXX", "CXXX"],
"difficulty": "medium"
},
{
"id": "q026",
"query": "吸烟环境的维护交接流程是怎样的?",
"query_type": "reasoning",
"relevant_docs": ["2.docx"],
"reference_answer": "维护交接需要按照维护标准中的规定,完成正式的交接手续,填写交接记录表,明确交接双方的责任和义务。",
"expected_keywords": ["交接", "手续", "记录", "责任", "维护标准"],
"difficulty": "medium"
},
{
"id": "q027",
"query": "吸烟设施的监督检查包括哪些内容?",
"query_type": "enumeration",
"relevant_docs": ["2.docx"],
"reference_answer": "监督检查是管理标准中的专门章节,包括对设施运行状态、卫生状况、设备完好程度、使用规范性等方面的检查。",
"expected_keywords": ["监督检查", "运行状态", "卫生", "设备", "规范性"],
"difficulty": "medium"
},
{
"id": "q028",
"query": "零售终端的分类体系是怎样的?从高到低怎么排列?",
"query_type": "enumeration",
"relevant_docs": ["3.docx"],
"reference_answer": "零售终端从高到低分为:直营终端、合作终端、加盟终端(星级加盟/普通加盟、现代终端A/B/C/D/E五个等级、普通终端含'五化'普通终端)。",
"expected_keywords": ["直营终端", "合作终端", "加盟终端", "现代终端", "普通终端", "A", "B", "C", "D", "E"],
"difficulty": "medium"
},
{
"id": "q029",
"query": "加盟终端使用什么品牌名称?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端使用'金丝利零售'Kingsley Retail品牌。",
"expected_keywords": ["金丝利零售", "Kingsley", "Retail", "品牌"],
"difficulty": "easy"
},
{
"id": "q030",
"query": "加盟终端的加盟协议期限是多长?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "首次加盟协议期限为三年,续签加盟协议期限也为三年。注意:智能终端设备合作协议和现代终端合作协议的期限是五年,不要混淆。",
"expected_keywords": ["三年", "加盟协议", "期限"],
"difficulty": "easy"
},
{
"id": "q031",
"query": "加盟终端评价的合格分数线是多少?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端运行评价满分为100分达标分为80分。一般现代终端评价同样为满分100分、达标分80分。",
"expected_keywords": ["100分", "80分", "达标", "评价"],
"difficulty": "easy"
},
{
"id": "q032",
"query": "星标加盟终端的评分标准是什么?城网和农网有什么区别?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "星标加盟终端评价标准城网要求评价周期内平均分95分及以上、单次得分不低于90分农网要求平均分90分及以上、单次得分不低于85分。两者都需要评价周期内积极配合市县两级公司开展各类活动。",
"expected_keywords": ["95分", "90分", "85分", "星标", "城网", "农网"],
"difficulty": "medium"
},
{
"id": "q033",
"query": "加盟终端之间的最小间距要求是多少?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端布局间步行距离应不低于300米。",
"expected_keywords": ["300米", "步行距离", "不低于"],
"difficulty": "easy"
},
{
"id": "q034",
"query": "加盟终端的卷烟展示面积要求是多少?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "卷烟展示面积不低于1.8平方米。",
"expected_keywords": ["1.8", "平方米", "展示面积", "不低于"],
"difficulty": "easy"
},
{
"id": "q035",
"query": "加盟终端的每日营业时间要求是多少?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端营业时间要求每天12小时以上。",
"expected_keywords": ["12小时", "营业时间", "每天"],
"difficulty": "easy"
},
{
"id": "q036",
"query": "加盟终端的'六个统一'是什么?",
"query_type": "enumeration",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端实行'六个统一'管理体系,涵盖品牌形象、服务标准、运营管理等方面的统一要求。",
"expected_keywords": ["六个统一", "品牌形象", "服务标准", "管理"],
"difficulty": "medium"
},
{
"id": "q037",
"query": "现代终端分为几个等级?分别是什么?",
"query_type": "enumeration",
"relevant_docs": ["3.docx"],
"reference_answer": "一般现代终端分为两层五类新现代终端A类、B类、C类和普通现代终端D类、E类。A类使用江苏烟草自采智能终端设备B类使用非自采智能终端设备C类使用简易智能终端设备D类通过数据接口自动上传E类通过客户手动上传。",
"expected_keywords": ["A类", "B类", "C类", "D类", "E类", "新现代", "普通现代"],
"difficulty": "easy"
},
{
"id": "q038",
"query": "终端运营管理中有哪些禁止行为?",
"query_type": "enumeration",
"relevant_docs": ["3.docx"],
"reference_answer": "终端运营管理中规定了10条禁止行为用于保护品牌价值和规范终端经营行为。",
"expected_keywords": ["10条", "禁止行为", "品牌价值", "经营"],
"difficulty": "medium"
},
{
"id": "q039",
"query": "加盟终端和合作终端有什么区别?",
"query_type": "comparison",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端使用'金丝利零售'品牌签订3年加盟协议实行'六个统一'管理。合作终端是另一种合作模式,与加盟终端在品牌使用、管理方式、投入标准等方面有所不同。",
"expected_keywords": ["金丝利零售", "加盟协议", "合作", "品牌", "管理方式"],
"difficulty": "medium"
},
{
"id": "q040",
"query": "终端评价不合格会怎么处理?",
"query_type": "reasoning",
"relevant_docs": ["3.docx"],
"reference_answer": "终端评价不合格将进入退出管理流程,根据评价结果进行降级或退出处理。退出管理定义了具体的降级和移除程序。",
"expected_keywords": ["退出管理", "降级", "退出", "评价", "不合格"],
"difficulty": "medium"
},
{
"id": "q041",
"query": "直营终端有什么特殊要求?",
"query_type": "reasoning",
"relevant_docs": ["3.docx"],
"reference_answer": "直营终端是由公司直接拥有和运营的终端,在品牌形象、服务标准、运营管理等方面执行最高标准,是终端体系中的最高层级。",
"expected_keywords": ["直营", "公司", "运营", "最高", "层级"],
"difficulty": "medium"
},
{
"id": "q042",
"query": "普通终端如何升级为现代终端?",
"query_type": "reasoning",
"relevant_docs": ["3.docx"],
"reference_answer": "普通终端可通过'五化'建设升级为五化普通终端,进而达到现代终端标准。五化是普通终端的升级路径,包括店面形象、经营管理、信息化等方面的提升。",
"expected_keywords": ["五化", "升级", "普通终端", "现代终端", "店面", "经营管理"],
"difficulty": "hard"
},
{
"id": "q043",
"query": "终端建设标准的附件有多少个?",
"query_type": "simple_fact",
"relevant_docs": ["3.docx"],
"reference_answer": "终端梯次化建设标准共有20个附件包括新现代终端客户申请书、现代终端合作协议、加盟终端合作协议、评价评分表、审批表、维护申请表等各类模板。",
"expected_keywords": ["20个", "附件", "申请书", "合作协议", "评价"],
"difficulty": "easy"
},
{
"id": "q044",
"query": "三峡工程有哪些综合效益?",
"query_type": "enumeration",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "三峡工程的综合效益包括防洪、发电、航运、水资源利用、生态环境保护等。三峡电站是世界上总装机容量最大的水电站。",
"expected_keywords": ["防洪", "发电", "航运", "水资源", "生态", "综合效益"],
"difficulty": "easy"
},
{
"id": "q045",
"query": "2022年三峡电站发电量为什么比往年低",
"query_type": "reasoning",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "2022年三峡电站年度发电量为787.90亿千瓦时为2012年以来最低值。主要原因是长江流域遭遇1961年以来最严重的夏秋连旱7-9月入库水量仅1007亿立方米。",
"expected_keywords": ["787.90", "干旱", "1961年", "夏秋连旱", "1007亿", "入库水量"],
"difficulty": "medium"
},
{
"id": "q046",
"query": "三峡工程在防洪方面发挥了什么作用?",
"query_type": "definition",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "三峡水库蓄水运行以来至2022年累计汛期拦洪总量2005.16亿立方米,有效应对了多次区域性大洪水,大大降低了中下游洪水位。",
"expected_keywords": ["2005.16亿", "拦洪", "汛期", "洪水位", "中下游"],
"difficulty": "medium"
},
{
"id": "q047",
"query": "三峡船闸和升船机2022年的运行情况对比如何",
"query_type": "comparison",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "2022年三峡船闸运行10400闸次通过船舶40641艘次货运量1.56亿吨通航率92.72%。升船机运行4470厢次通过船舶4506艘次旅客54416人次。",
"expected_keywords": ["10400", "40641", "1.56亿吨", "92.72%", "4470", "54416"],
"difficulty": "medium"
},
{
"id": "q048",
"query": "三峡工程对节能减排有什么贡献?",
"query_type": "reasoning",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "截至2022年底三峡电站累计发出清洁电力相当于节约标准煤4.85亿吨减少二氧化碳排放12.65亿吨。",
"expected_keywords": ["4.85亿吨", "标准煤", "12.65亿吨", "二氧化碳", "清洁电力"],
"difficulty": "medium"
},
{
"id": "q049",
"query": "2022年三峡水库的泥沙淤积情况如何排沙比是多少",
"query_type": "table_data",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "2022年三峡水库入库悬移质输沙量0.136亿吨出库0.026亿吨库区淤积0.110亿吨排沙比19.3%。",
"expected_keywords": ["0.136亿吨", "0.026亿吨", "0.110亿吨", "19.3%", "排沙比"],
"difficulty": "medium"
},
{
"id": "q050",
"query": "2022年三峡水库蓄水目标完成了吗为什么",
"query_type": "reasoning",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "2022年三峡水库库水位最高蓄至160.04米未完成175米蓄水目标。原因是长江流域遭遇严重干旱入库水量减少同时为应对下游旱情多次补水。",
"expected_keywords": ["160.04米", "175米", "未完成", "干旱", "补水"],
"difficulty": "hard"
},
{
"id": "q051",
"query": "2022年三峡水库为什么频繁向下游补水补了多少",
"query_type": "reasoning",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "2022年长江流域遭遇严重干旱三峡水库频繁向下游补水是为缓解下游旱情。枯水季节累计补水158天补水总量217.76亿立方米。",
"expected_keywords": ["158天", "217.76亿", "旱情", "枯水季节", "补水"],
"difficulty": "medium"
},
{
"id": "q052",
"query": "三峡工程2022年进行了哪些生态调度试验",
"query_type": "enumeration",
"relevant_docs": ["三峡公报_1-15页.pdf"],
"reference_answer": "2022年进行了三类生态调度1.针对四大家鱼等产漂流性卵鱼类繁殖的水文生态调度试验2.针对库区产黏沉性卵鱼类自然繁殖的生态调度试验3.库尾减淤调度试验。",
"expected_keywords": ["四大家鱼", "漂流性卵", "黏沉性卵", "减淤", "生态调度"],
"difficulty": "medium"
},
{
"id": "q053",
"query": "三份工作规范文档分别涉及什么领域?各自适用范围是什么?",
"query_type": "cross_doc",
"relevant_docs": ["1.docx", "2.docx", "3.docx"],
"reference_answer": "1.docx是货源投放工作规范适用于烟草产品的供应分配管理。2.docx是文明吸烟环境建设/管理/维护标准适用于吸烟设施的建设与运维。3.docx是零售终端梯次化建设标准适用于卷烟零售终端的分级管理。三份文档均属于江苏省烟草行业规范。",
"expected_keywords": ["货源投放", "吸烟环境", "零售终端", "江苏省", "烟草"],
"difficulty": "hard"
},
{
"id": "q054",
"query": "三份文档中的分类体系有什么相似之处?",
"query_type": "cross_doc",
"relevant_docs": ["1.docx", "2.docx", "3.docx"],
"reference_answer": "三份文档都采用了分层分类的管理方法:货源投放将品规分为紧俏/均衡/宽松三类,客户按档级分类;吸烟环境将设施分为室/区/点三类并按A/B/C区域分类零售终端分为普通/现代/加盟/合作/直营五个层级。",
"expected_keywords": ["分类", "分层", "品规", "设施", "终端", "层级"],
"difficulty": "hard"
},
{
"id": "q055",
"query": "货源投放规范和终端建设标准在评价体系上有什么共同点?",
"query_type": "cross_doc",
"relevant_docs": ["1.docx", "3.docx"],
"reference_answer": "两者都采用了量化评分体系货源投放使用双层双级指标体系评价市场状态终端建设使用100分制评价终端达标情况。都设置了明确的评分阈值和等级划分标准。",
"expected_keywords": ["量化", "评分", "指标体系", "阈值", "等级"],
"difficulty": "hard"
},
{
"id": "q056",
"query": "我们公司食堂在哪里?",
"query_type": "negative",
"relevant_docs": [],
"reference_answer": "知识库中没有关于公司食堂的信息。现有文档主要涉及货源投放工作规范、文明吸烟环境标准和零售终端建设标准。",
"expected_keywords": ["未找到", "未"],
"difficulty": "easy"
},
{
"id": "q057",
"query": "公司的股票代码是什么?",
"query_type": "negative",
"relevant_docs": [],
"reference_answer": "知识库中没有关于股票代码的信息。现有文档主要涉及烟草行业的工作规范和建设标准。",
"expected_keywords": ["未找到", "未"],
"difficulty": "easy"
},
{
"id": "q058",
"query": "如何申请年休假?需要哪些流程?",
"query_type": "negative",
"relevant_docs": [],
"reference_answer": "知识库中没有关于年休假申请流程的信息。现有文档涉及的是货源投放、吸烟环境建设和零售终端管理。",
"expected_keywords": ["未找到", "未"],
"difficulty": "easy"
},
{
"id": "q059",
"query": "紧俏品规的上限比例是多少?",
"query_type": "paraphrase",
"relevant_docs": ["1.docx"],
"reference_answer": "紧俏品规数不得超过地市级公司所经营品规数的20%。",
"expected_keywords": ["20%", "不得超过"],
"difficulty": "easy",
"paraphrase_of": "q004"
},
{
"id": "q060",
"query": "吸烟环境的七种功能分别是什么?",
"query_type": "paraphrase",
"relevant_docs": ["2.docx"],
"reference_answer": "SUCCESS功能框架包含7个功能S-Smoking吸烟、U-Utility实用、C-Convenience便利、C-Comfortable舒适、E-Experience体验、S-Safe安全、S-Survey调研。",
"expected_keywords": ["Smoking", "Utility", "Convenience", "Comfortable", "Experience", "Safe", "Survey"],
"difficulty": "medium",
"paraphrase_of": "q018"
},
{
"id": "q061",
"query": "加盟门店之间最少要隔多远?",
"query_type": "paraphrase",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端之间的最小间距为300米。",
"expected_keywords": ["300米", "间距"],
"difficulty": "easy",
"paraphrase_of": "q033"
},
{
"id": "q062",
"query": "金丝利零售门店每天最少营业多长时间?",
"query_type": "paraphrase",
"relevant_docs": ["3.docx"],
"reference_answer": "加盟终端每日营业时间不低于12小时。",
"expected_keywords": ["12小时", "营业"],
"difficulty": "easy",
"paraphrase_of": "q035"
}
],
"statistics": {
"total_queries": 62,
"by_category": {
"simple_fact": 15,
"enumeration": 13,
"reasoning": 11,
"comparison": 7,
"paraphrase": 4,
"definition": 3,
"table_data": 3,
"cross_doc": 3,
"negative": 3
},
"by_difficulty": {
"easy": 22,
"medium": 30,
"hard": 10
},
"by_document": {
"1.docx": 21,
"2.docx": 13,
"3.docx": 21,
"三峡公报_1-15页.pdf": 9,
"negative(无相关文档)": 3,
"cross_doc(多文档)": 3,
"paraphrase(改写)": 4
}
}
}