data: drop login_required tasks (220 -> 217)

LexBench-Browser/task.jsonl

@@ -171,9 +171,6 @@
 {"id": 304, "query": "在经济观察网搜索\"房地产市场\"，找到最新且标题中包含\"房地产市场\"字样的文章，记录文章标题、发布时间和来源。", "task_type": "T1", "domain": "finance_gaming", "difficulty": "easy", "login_required": false, "login_type": "", "risk_control": false, "risk_control_types": [], "target_website": "www.eeo.com.cn", "reasoning_type": "multi_step", "language": "zh", "website_region": "zh", "reference_answer": {"steps": ["访问经济观察网 www.eeo.com.cn", "在搜索框输入\"房地产市场\"", "点击搜索", "浏览搜索结果列表", "注意：该网站搜索结果无按时间排序功能，需手动浏览找到最新且标题含\"房地产市场\"的文章", "找到符合条件的文章后点击进入", "记录文章标题、发布时间和来源"], "key_points": ["无需登录即可搜索", "经济观察网搜索结果没有按时间排序按钮，这是该网站的设计特点", "必须找标题中包含\"房地产市场\"字样的文章，不能选标题无关的文章", "需要在搜索结果中手动浏览对比找到最新的一篇"], "common_mistakes": ["标题不匹配：选取了标题中没有\"房地产市场\"字样的文章", "未找到最新的：没有对比多篇文章的发布时间", "搜索结果页无排序功能：不要花时间寻找不存在的排序按钮", "信息不完整：遗漏标题、发布时间或来源"], "scoring": {"total": 100, "items": [{"name": "成功搜索", "score": 15, "description": "正确搜索房地产市场"}, {"name": "找到标题匹配的文章", "score": 30, "description": "找到标题中包含\"房地产市场\"的文章"}, {"name": "找到最新的一篇", "score": 25, "description": "在多个结果中找到发布时间最近的"}, {"name": "记录标题", "score": 10, "description": "标题准确"}, {"name": "记录发布时间", "score": 10, "description": "发布时间正确"}, {"name": "记录来源", "score": 10, "description": "来源信息正确"}]}}, "score_threshold": 65, "robustness_tags": ["chinese_rendering", "filter_sort", "login_popup", "realtime_data"]}
 {"id": 306, "query": "在21世纪经济报道搜索\"能源转型\"，找到最新发布的一篇相关文章，记录文章标题、发布时间和来源。", "task_type": "T1", "domain": "finance_gaming", "difficulty": "easy", "login_required": false, "login_type": "", "risk_control": false, "risk_control_types": [], "target_website": "www.21jingji.com", "reasoning_type": "single_step", "language": "zh", "website_region": "zh", "reference_answer": {"steps": ["访问21世纪经济报道 www.21jingji.com", "在搜索框输入\"能源转型\"", "点击搜索", "浏览搜索结果列表", "找到最新发布的一篇相关文章", "点击进入文章详情页", "记录文章标题、发布时间和来源"], "key_points": ["无需登录即可搜索和浏览", "搜索结果默认按相关性排序，需注意对比发布时间找最新的", "21世纪经济报道网页版文章不显示阅读量", "来源通常显示在文章详情页的标题下方"], "common_mistakes": ["未对比发布时间：直接选了第一个结果，但不一定是最新的", "阅读量不可获取：该网站不显示阅读量，不要花时间寻找", "信息不完整：需要同时记录标题、发布时间和来源三项"], "scoring": {"total": 100, "items": [{"name": "成功搜索", "score": 15, "description": "正确搜索能源转型"}, {"name": "找到最新文章", "score": 30, "description": "找到发布时间最近的相关文章"}, {"name": "记录标题", "score": 20, "description": "标题准确"}, {"name": "记录发布时间", "score": 20, "description": "发布时间正确"}, {"name": "记录来源", "score": 15, "description": "来源信息正确"}]}}, "score_threshold": 65, "robustness_tags": ["chinese_rendering", "filter_sort", "login_popup", "realtime_data"]}
 {"id": 308, "query": "在证券时报网进入\"股市\"频道，找到当前头条新闻，记录其标题和核心内容。然后利用网站的搜索功能，搜索该新闻涉及的关键词，找到至少2篇相关背景报道，综合分析该头条新闻事件的成因。", "task_type": "T2", "domain": "finance_gaming", "difficulty": "hard", "login_required": false, "login_type": "", "risk_control": false, "risk_control_types": [], "target_website": "www.stcn.com", "reasoning_type": "deep_analysis", "language": "zh", "website_region": "zh", "reference_answer": {"steps": ["访问证券时报网 www.stcn.com", "在首页导航栏找到\"股市\"频道入口并点击进入", "在股市频道页面找到当前头条新闻（通常是页面顶部最显眼的位置）", "点击进入头条新闻详情页", "阅读文章内容，记录标题和核心内容", "从文章中提取关键词（如涉及的行业、政策、事件名称等）", "返回首页或找到搜索框", "输入提取的关键词进行搜索", "从搜索结果中找到至少2篇与头条新闻相关的背景报道", "分别点击进入这些文章，阅读其内容", "综合头条新闻和背景���道的信息，分析头条新闻事件的成因", "整理并输出：头条标题、核心内容、搜索关键词、相关报道信息、成因分析"], "key_points": ["无需登录即可浏览股市频道和搜索文章", "必须通过导航栏进入股市频道，不能直接搜索", "头条新闻通常在频道页面最显眼的位置（大图或置顶）", "agent 需要自主理解头条内容并决定搜索什么关键词，这是任务的核心难点", "相关背景报道必须与头条新闻有实际关联，不能随意选取", "成因分析需要基于多篇文章的实际内容，不能凭空推测"], "common_mistakes": ["频道导航：未进入股市频道，直接在首页取文章", "头条识别：选取了非头条位置的文章", "关键词提取不当：搜索了过于宽泛或不相关的关键词，导致找不到有关联的报道", "相关性不足：找到的背景报道与头条新闻无实际关联", "分析流于表面：仅罗列文章内容而未真正分析事件成因", "报道数量不足：未找到至少2篇相关背景报道"], "scoring": {"total": 100, "items": [{"name": "进入股市频道", "score": 10, "description": "通过导航栏成功进入股市频道"}, {"name": "找到头条新闻", "score": 10, "description": "定位到股市频道当前头条新闻"}, {"name": "记录头条标题和核心内容", "score": 10, "description": "准确记录头条新闻的标题和核心内容"}, {"name": "提取关键词并搜索", "score": 15, "description": "agent 自主从头条中提取合理的关键词并执行搜索"}, {"name": "找到相关背景报道", "score": 15, "description": "从搜索结果中找到至少2篇与头条新闻有实际关联的报道"}, {"name": "阅读背景报道内容", "score": 10, "description": "进入相关报道详情页并阅读内容"}, {"name": "成因分析质量", "score": 30, "description": "基于头条和背景报道的实际内容，给出合理、有依据的成因分析"}]}}, "score_threshold": 55, "robustness_tags": ["chinese_rendering", "deep_navigation", "login_popup"]}
-{"id": 3001, "query": "Navigate to https://xueqiu.com and wait 30 seconds for the user to manually log in and complete any CAPTCHA if required (the browser window is visible). After logging in, find a user who has published discussion posts about stocks, then collect all comments from the first 10 posts published by this user. Save all collected data to JSON files. After collecting all comments, confirm that the data has been saved.", "task_type": "T1", "domain": "finance_gaming", "difficulty": "medium", "login_required": true, "login_type": "account_password", "risk_control": false, "risk_control_types": [], "target_website": "xueqiu.com", "reasoning_type": "multi_step", "language": "en", "website_region": "zh", "reference_answer": {"steps": ["访问雪球网站 https://xueqiu.com", "等待30秒供用户手动登录并完成验证码", "登录后浏览首页或热门讨论区", "寻找发布过股票讨论帖的用户（通过帖子内容或用户标签判断）", "进入该用户的个人主页", "定位用户发布的讨论帖列表", "遍历前10个帖子，逐一进入每个帖子详情页", "对每个帖子，滚动加载并采集所有评论（用户名、评论内容、时间、点赞数等）", "将采集的数据结构化存储为JSON文件", "确认文件保存成功"], "key_points": ["必须等待30秒让用户完成登录和验证", "需要自主判断并选择合适的用户（有股票讨论帖）", "必须采集前10个帖子的所有评论，不能遗漏", "数据需要结构化存储为JSON格式"], "common_mistakes": ["未等待足够时间导致登录未完成", "选择的用户没有股票相关讨论", "只采集部分评论，未处理分页加载", "数据格式不规范或未保存文件"], "scoring": {"total": 100, "items": [{"name": "成功等待并识别登录完成", "score": 5, "description": "正确等待30秒并确认用户已登录"}, {"name": "找到合适的用户", "score": 10, "description": "选择的用户确实发布过股票讨论帖"}, {"name": "定位前10个帖子", "score": 10, "description": "正确识别用户的前10个帖子"}, {"name": "遍历所有帖子", "score": 10, "description": "成功访问全部10个帖子详情页"}, {"name": "采集评论完整性", "score": 35, "description": "每个帖子的评论都完整采集（处理分页/滚动加载）"}, {"name": "数据结构化", "score": 20, "description": "数据格式规范（JSON），包含必要字段（用户名、内容、时间等）"}, {"name": "保存文件并确认", "score": 10, "description": "文件成功保存且路径可访问"}]}}, "score_threshold": 70, "robustness_tags": ["chinese_rendering", "cross_language", "data_extraction", "deep_navigation", "lazy_load_scroll", "login_popup"]}
-{"id": 3002, "query": "⚠️ IMPORTANT: Navigate to https://www.bilibili.com and IMMEDIATELY wait 30 seconds for the user to manually log in (the browser window is visible). DO NOT proceed with any other actions until the 30-second wait is complete. After the wait period, search for \"iphone17\" videos on Bilibili, find the video with the highest play count, and collect 100 comments from that video.", "task_type": "T1", "domain": "video_platform", "difficulty": "medium", "login_required": true, "login_type": "account_password", "risk_control": false, "risk_control_types": [], "target_website": "www.bilibili.com", "reasoning_type": "multi_step", "language": "en", "website_region": "zh", "reference_answer": {"steps": ["访问B站 https://www.bilibili.com", "立即等待30秒，不进行任何操作，供用户手动登录", "确认30秒等待完成后，定位搜索框", "在搜索框输入关键词 iphone17", "提交搜索并等待结果页加载", "选择按最多播放排序", "找到播放量最高的视频", "点击进入该视频详情页", "滚动到评论区，逐步加载评论（可能需要多次滚动）", "采集100条评论（包括用户名、评论内容、点赞数、回复数等）", "将数据保存为JSON文件"], "key_points": ["必须严格等待30秒后再操作", "必须基于播放量选择视频，不能随意选择", "必须采集完整的100条评论", "评论数据需包含关键信息（用户名、内容、点赞数等）"], "common_mistakes": ["未等待30秒就开始操作", "按其他排序选择视频（如综合排序而非播放量）", "评论采集不足100条或采集过程中断", "未处理评论分页/懒加载"], "scoring": {"total": 100, "items": [{"name": "正确执行30秒等待", "score": 5, "description": "严格等待30秒供用户登录"}, {"name": "成功搜索iphone17", "score": 10, "description": "搜索关键词正确提交"}, {"name": "识别播放量最高视频", "score": 15, "description": "正确比较播放量并选择最高的"}, {"name": "进入视频详情页", "score": 5, "description": "成功访问视频页面"}, {"name": "采集100条评论", "score": 45, "description": "完整采集100条评论（处理分页加载）"}, {"name": "数据完整性", "score": 15, "description": "评论包含视频BV号、用户名、内容、点赞数、回复数"}, {"name": "保存为JSON文件", "score": 5, "description": "数据成功保存且格式正确"}]}}, "score_threshold": 70, "robustness_tags": ["ad_overlay", "chinese_rendering", "cross_language", "data_extraction", "filter_sort", "iframe_embed", "lazy_load_scroll", "login_popup"]}
-{"id": 3003, "query": "⚠️ IMPORTANT: Navigate to https://www.bilibili.com and IMMEDIATELY wait 30 seconds for the user to manually log in (the browser window is visible). DO NOT proceed with any other actions until the 30-second wait is complete. After the wait period, search for \"科技\" videos on Bilibili, find the video with the highest play count, and collect 100 comments from that video. Then search for \"数码\" videos on Bilibili, find the video with the highest play count, and collect 100 comments from that video. For each comment, provide the video BV number, commenter's username, comment content, like count, and reply count. Save all collected data to JSON files.", "task_type": "T1", "domain": "video_platform", "difficulty": "medium", "login_required": true, "login_type": "account_password", "risk_control": false, "risk_control_types": [], "target_website": "www.bilibili.com", "reasoning_type": "multi_step", "language": "zh", "website_region": "zh", "reference_answer": {"steps": ["访问B站 https://www.bilibili.com", "立即等待30秒，供用户手动登录", "【第一轮】搜索 科技 关键词", "选择按最多播放排序", "选择播放量最高的科技视频", "进入视频详情页，采集100条评论（包含BV号、用户名、内容、点赞数、回复数）", "【第二轮】返回搜索页或首页", "搜索 数码 关键词", "选择按最多播放排序", "选择播放量最高的数码视频", "进入视频详情页，采集100条评论（包含BV号、用户名、内容、点赞数、回复数）", "将两个视频的评论数据分别或合并保存为JSON文件"], "key_points": ["必须等待30秒供用户登录", "需要完成两轮搜索和采集，每轮100条评论", "每条评论必须包含5个字段：BV号、用户名、内容、点赞数、回复数", "两个视频的数据需要区分标识"], "common_mistakes": ["只完成一个关键词的采集", "评论数据缺少必需字段", "未正确识别播放量最高的视频", "评论数量不足200条（每个视频100条）"], "scoring": {"total": 100, "items": [{"name": "正确执行30秒等待", "score": 5, "description": "严格等待30秒供用户登录"}, {"name": "完成科技视频搜索", "score": 10, "description": "成功搜索并找到播放量最高的科技视频"}, {"name": "采集科技视频100条评论", "score": 25, "description": "完整采集科技视频的100条评论"}, {"name": "完成数码视频搜索", "score": 10, "description": "成功搜索并找到播放量最高的数码视频"}, {"name": "采集数码视频100条评论", "score": 25, "description": "完整采集数码视频的100条评论"}, {"name": "评论字段完整性", "score": 20, "description": "每条评论包含BV号、用户名、内容、点赞数、回复数"}, {"name": "保存为JSON文件", "score": 5, "description": "数据成功保存且格式正确"}]}}, "score_threshold": 70, "robustness_tags": ["ad_overlay", "chinese_rendering", "data_extraction", "deep_navigation", "filter_sort", "iframe_embed", "login_popup"]}
 {"id": 3004, "query": "Get the current stock data for 20 stocks: BYD (比亚迪, 002594), CATL (宁德时代, 300750), Kweichow Moutai (贵州茅台, 600519), Wuliangye (五粮液, 000858), Ping An (中国平安, 601318), China Merchants Bank (招商银行, 600036), Midea Group (美的集团, 000333), Gree Electric (格力电器, 000651), Sany Heavy Industry (三一重工, 600031), Hikvision (海康威视, 002415), Luxshare Precision (立讯精密, 002475), BOE (京东方A, 000725), TCL Technology (TCL科技, 000100), SMIC (中芯国际, 688981), Unigroup Guoxin (紫光国微, 002049), Will Semiconductor (韦尔股份, 603501), Inovance Technology (汇川技术, 300124), Sungrow Power (阳光电源, 300274), LONGi Green Energy (隆基绿能, 601012), Tongwei (通威股份, 600438). For each stock, provide the stock code, name, current price, price change, price change percentage, highest price, lowest price, trading volume, and market value. Organize all data in a structured table format.", "task_type": "T1", "domain": "finance_gaming", "difficulty": "easy", "login_required": false, "login_type": "", "risk_control": false, "risk_control_types": [], "target_website": "www.eastmoney.com", "reasoning_type": "single_step", "language": "zh", "website_region": "zh", "reference_answer": {"steps": ["访问东方财富网 https://www.eastmoney.com", "逐个搜索20只股票（可通过股票代码或名称）", "对每只股票，访问详情页或通过API获取数据", "提取9个字段：股票代码、名称、现价、涨跌额、涨跌幅、最高价、最低价、成交量、市值", "将20只股票的数据整理为结构化表格", "确保数据格式统一（价格保留小数、百分比格式等）"], "key_points": ["必须查询全部20只股票", "每只股票必须包含9个指定字段", "数据需要实时准确（当前交易数据）", "最终以表格形式组织数据"], "common_mistakes": ["股票代码输入错误或查询失败", "数据字段不全（缺少市值、成交量等）", "数据格式不统一（单位混乱）", "未完成全部20只股票的查询"], "scoring": {"total": 100, "items": [{"name": "访问东方财富网", "score": 5, "description": "成功访问网站"}, {"name": "查询20只股票", "score": 20, "description": "成功查询全部20只股票（每只1分）"}, {"name": "数据字段完整性", "score": 45, "description": "每只股票包含9个指定字段（代码、名称、现价、涨跌额、涨跌幅、最高、最低、成交量、市值）"}, {"name": "数据准确性", "score": 15, "description": "数据为实时/当日数据，数值合理"}, {"name": "表格格式组织", "score": 15, "description": "数据以结构化表格形式呈现，格式统一"}]}}, "score_threshold": 70, "robustness_tags": ["chinese_rendering", "data_extraction", "realtime_data"]}
 {"id": 3005, "query": "Query train tickets for 10 popular city pairs on Ctrip: (1) Beijing to Shanghai, (2) Shanghai to Hangzhou, (3) Hangzhou to Guangzhou, (4) Guangzhou to Shenzhen, (5) Beijing to Xi'an, (6) Xi'an to Chengdu, (7) Chengdu to Chongqing, (8) Shanghai to Nanjing, (9) Nanjing to Suzhou, (10) Suzhou to Hangzhou. For each city pair, query tickets for tomorrow and save all collected data to JSON files. After collecting all information, create a summary table showing each city pair and the total number of trains found. Confirm that the data has been saved.", "task_type": "T2", "domain": "social_lifestyle", "difficulty": "easy", "login_required": false, "login_type": "", "risk_control": false, "risk_control_types": [], "target_website": "trains.ctrip.com", "reasoning_type": "single_step", "language": "en", "website_region": "zh", "reference_answer": {"steps": ["访问携程火车票查询页面 https://trains.ctrip.com", "逐个查询10对城市的火车票", "对每对城市：选择出发城市、到达城市、明天日期", "提交查询，等待结果加载", "采集所有车次数据（车次号、出发时间、到达时间、历时、票价等）", "将每对城市的数据保存到JSON文件", "汇总统计：创建表格显示10对城市及对应车次总数", "确认所有数据文件已保存"], "key_points": ["必须查询全部10对城市", "查询日期必须是明天", "需要采集完整的车次列表，不能只采集部分", "最终需要汇总表格统计车次数量"], "common_mistakes": ["城市名称输入错误导致查询失败", "日期选择错误（不是明天）", "只采集首屏车次，未处理滚动加载", "未创建汇总统计表"], "scoring": {"total": 100, "items": [{"name": "访问携程火车票页面", "score": 5, "description": "成功访问查询页面"}, {"name": "查询10对城市", "score": 20, "description": "成功查询全部10对城市（每���2分）"}, {"name": "日期选择正确", "score": 10, "description": "所有查询日期为明天"}, {"name": "数据采集完整性", "score": 40, "description": "每对城市的所有车次数据完整采集"}, {"name": "保存为JSON文件", "score": 10, "description": "数据成功保存为JSON格式"}, {"name": "创建汇总表格", "score": 10, "description": "统计表显示10对城市和车次总数"}, {"name": "确认保存完成", "score": 5, "description": "明确确认数据已保存"}]}}, "score_threshold": 70, "robustness_tags": ["ad_overlay", "chinese_rendering", "cross_language", "data_extraction", "login_popup"]}
 {"id": 3006, "query": "Search for MODs on 3DM Mod站 for 5 popular games: (1) Elden Ring (艾尔登法环), (2) The Witcher 3 (巫师3), (3) Cyberpunk 2077 (赛博朋克2077), (4) Baldur's Gate 3 (博德之门3), (5) Starfield (星空). For each game, find the top 3 MODs by download count, then get the detailed information for each MOD including: mod name, download count, update time, and description (first 200 characters). Save all collected data to JSON files. After collecting all MOD details, confirm that the data has been saved.", "task_type": "T1", "domain": "finance_gaming", "difficulty": "easy", "login_required": false, "login_type": "", "risk_control": false, "risk_control_types": [], "target_website": "mod.3dmgame.com", "reasoning_type": "multi_step", "language": "zh", "website_region": "zh", "reference_answer": {"steps": ["访问3DM Mod站 https://mod.3dmgame.com", "逐个搜索5款游戏", "对每款游戏：按下载量排序，定位前3个MOD", "点击进入每个MOD的详情页", "采集MOD详细信息：名称、下载量、更新时间、描述（前200字符）", "共采集15个MOD的详细信息（5款游戏 × 3个MOD）", "将数据保存为JSON文件", "确认文件保存成功"], "key_points": ["必须搜索全部5款游戏", "每款游戏必须找到下载量最高的3个MOD", "必须进入详情页获取完整信息", "描述字段需截取前200字符"], "common_mistakes": ["游戏名称搜索错误（中英文混淆）", "排序方式错误（未按下载量排序）", "只采集列表信息，未进入详情页", "描述字段未截取或超出200字符"], "scoring": {"total": 100, "items": [{"name": "成功搜索5款游戏", "score": 15, "description": "全部5款游戏搜索成功（每款3分）"}, {"name": "按下载量排序", "score": 10, "description": "正确使用下载量排序功能"}, {"name": "定位前3个MOD", "score": 10, "description": "每款游戏准确定位下载量前3的MOD"}, {"name": "采集15个MOD详情", "score": 45, "description": "成功采集全部15个MOD的详细信息（每个3分）"}, {"name": "数据字段完整", "score": 10, "description": "每个MOD包含4个字段：名称、下载量、更新时间、描述"}, {"name": "描述截取正确", "score": 5, "description": "描述字段为前200字符"}, {"name": "保存并确认", "score": 5, "description": "数据保存为JSON且确认成功"}]}}, "score_threshold": 70, "robustness_tags": ["chinese_rendering", "filter_sort", "login_popup"]}