| { |
| "task_id": "consumerfinance_013", |
| "domain": "CFPB_COMPLAINT_DATABASE", |
| "autonomy_type": "ordered table", |
| "oracle_output_cardinality": 2, |
| "instruction": "我正在撰写一份加密货币投诉“重灾区”分析报告（crypto-complaint concentration memo），为了核实一个固定三年窗口期内的投诉高峰是否仅由少数头部平台拉高，需要你协助我通过 CFPB 投诉数据库按步骤排查并提取相关数据。\n首先，我们需要界定分析的底层数据范围。请先根据 CFPB 当前投诉表单（complaint form）的产品分类口径，确认虚拟货币“Virtual currency”目前归属的产品大类（Product bucket）。在确定确切归属后，需将时间范围严格锁定在 2019-08-26 至 2022-08-26 之间，且仅提取该产品大类下“Virtual currency”子类的投诉记录，以此形成本次分析的初始群体（cohort）。\n其次，我们需要在这个初始群体中推演并锁定两个特征峰值月。第一步，需按月统计整体投诉量，锁定该初始群体在未移除任何公司时的原始峰值月（AllCompanies）。第二步，需精准识别出上述原始峰值月中投诉量最高的前两家公司，并将这两家公司在整个三年初始群体中的所有投诉记录彻底整体移除（不是仅移除单月记录），随后基于这批残余数据重新按月统计，沉淀出剔除头部影响后的残余峰值月（ResidualAfterTop2）。\n再次，针对上述锁定的这两个特征月份，我们需要分别提取详尽的数据。对于这两条投诉簇，除必须获取该月的总投诉量外，还需分别查实被移除的那两家头部巨头公司名称，以及它们在“原始峰值月”中的合计投诉量。同时，需明确该特征月份内投诉量最高的州及其投诉量、该州内被投诉最多的公司及其投诉规模，以及这家公司在该州、该特征月范围内收到的最早一条投诉的完整细节（包含投诉编号、日期、Issue 和 Sub-issue）。\n最后，在完成上述所有维度的排查与信息梳理后，若将提取到的这两组核心参数，严格按照 AllCompanies 在前、ResidualAfterTop2 在后的顺序，拼接成没有任何空格的格式化字符串（每行的字段拼接顺序务必固定为：<Cluster>|<Month>|<MonthComplaints>|<ExcludedTop2Companies>|<ExcludedTop2PeakMonthComplaints>|<TopState>|<StateComplaints>|<TopCompany>|<CompanyComplaints>|<EarliestComplaintID>|<EarliestDate>|<Issue>|<SubIssue>），最终生成的这组数据记录内容是什么？为保证答案唯一：两条记录按 AllCompanies、ResidualAfterTop2 的顺序输出；ExcludedTop2Companies 按原始峰值月投诉量从高到低排列；如任何排序指标出现并列，则按字段值字母序升序排列；若最早投诉日期并列，则取 Complaint ID 数值最小者。", |
| "start_url": "https://www.consumerfinance.gov/data-research/consumer-complaints/", |
| "output_format": "严格按照 AllCompanies 在前、ResidualAfterTop2 在后的顺序，拼接成没有任何空格的格式化字符串。每行的字段拼接顺序务必固定为：<Cluster>|<Month>|<MonthComplaints>|<ExcludedTop2Companies>|<ExcludedTop2PeakMonthComplaints>|<TopState>|<StateComplaints>|<TopCompany>|<CompanyComplaints>|<EarliestComplaintID>|<EarliestDate>|<Issue>|<SubIssue>。", |
| "oracle_answer": "AllCompanies|2021-05|336|Coinbase, Inc.+BAM Management US Holdings Inc.|226|CA|58|Coinbase, Inc.|21|4341451|2021-05-01|Money was not available when promised|NONE,ResidualAfterTop2|2020-06|180|Coinbase, Inc.+BAM Management US Holdings Inc.|226|OH|28|PNC Bank N.A.|22|3687376|2020-06-07|Other transaction problem|NONE", |
| "metadata": { |
| "State-Gated Retrieval": [ |
| "只看当前 product bucket 下 Virtual currency 子类、接收日期在 2019-08-26 至 2022-08-26 的 Complaint Database 记录。", |
| "先确定 AllCompanies 原始峰值月，再识别该峰值月中投诉量最高的前两家公司。", |
| "这两家公司必须从整个三年 cohort 中整体移除，然后基于残余 cohort 重新计算 residual peak month 及所有 downstream 字段。" |
| ], |
| "dependency_type": "Data + Control", |
| "intra_chain": true, |
| "inter_chain": true, |
| "data_dependency": [ |
| "current complaint-form taxonomy yields the product bucket for Virtual currency", |
| "complaint records within the fixed three-year cohort yield the all-company peak month, per-month top companies, and residual-cohort recomputations after company removal", |
| "final rows are produced only after the top-2 companies from the original peak month are removed from the full cohort and downstream aggregates are computed on the resulting cohort" |
| ], |
| "control_dependency": [ |
| "peak-month top companies must be removed from the entire cohort, not only from the original peak month", |
| "residual peak detection must be computed from the rewritten cohort rather than by local patching of the original month", |
| "downstream state / company / earliest-complaint fields are valid only after the global company-removal step is locked" |
| ], |
| "freeze": { |
| "historical_window": "Complaint Database records received from 2019-08-26 through 2022-08-26 under the current product-bucket mapping for Virtual currency" |
| }, |
| "answer_type": "multi-row ordered table" |
| }, |
| "rubric": { |
| "inclusion_conditions": [ |
| "只看当前 product bucket 下 Virtual currency 子类、接收日期在 2019-08-26 至 2022-08-26 的 Complaint Database 记录。", |
| "先确定 AllCompanies 原始峰值月，再识别该峰值月中投诉量最高的前两家公司。", |
| "这两家公司必须从整个三年 cohort 中整体移除，然后基于残余 cohort 重新计算 residual peak month 及所有 downstream 字段。" |
| ], |
| "exclusion_conditions": [ |
| "排除只从原始峰值月局部删除 top-2 公司记录的结果。", |
| "排除没有把 top-2 company set 回写到整个 cohort 后再重算 residual peak 的结果。", |
| "排除基于未重算残余 cohort 的 downstream 州、公司或 earliest complaint 字段。" |
| ], |
| "normalization": { |
| "field_separator": "|", |
| "record_separator": ",", |
| "schema": [ |
| "Cluster", |
| "Month", |
| "MonthComplaints", |
| "ExcludedTop2Companies", |
| "ExcludedTop2PeakMonthComplaints", |
| "TopState", |
| "StateComplaints", |
| "TopCompany", |
| "CompanyComplaints", |
| "EarliestComplaintID", |
| "EarliestDate", |
| "Issue", |
| "SubIssue" |
| ], |
| "cluster_order": [ |
| "AllCompanies", |
| "ResidualAfterTop2" |
| ], |
| "date_format": { |
| "Month": "YYYY-MM", |
| "EarliestDate": "YYYY-MM-DD" |
| }, |
| "ExcludedTop2Companies": "join the two excluded company names with `+` in source-peak complaint-count descending order, breaking ties alphabetically", |
| "TopState": "use the uppercase two-letter state abbreviation", |
| "TopCompany": "use the company display name as shown in the source data", |
| "tie_breaks": "for the earliest complaint use EarliestDate ascending then Complaint ID ascending", |
| "SubIssue": "emit `NONE` when the complaint row has no sub-issue value" |
| } |
| } |
| } |
|
|
