Nachammai41/underserved-persona_conditioned-fraud-v4

这是一个包含20,300条引用基础的合成欺诈叙述的数据集，专注于美国金融系统中服务不足的四种原型：汇款、零工工人、无银行账户者和ITIN持有者。数据集全面覆盖了25种FinCEN类型代码，并提供了详细的统计信息和多种配置选项。此外，还提供了一个单独的链式思维推理数据集，用于支持SFT、LLM-as-judge蒸馏或CoT数据增强。数据集支持多种语言，包括英语、西班牙语、越南语等，并包含详细的引用和许可信息。

A 20,300-row citation-grounded synthetic fraud-narrative dataset focusing on four underserved US financial-system archetypes: remittance, gig_worker, unbanked, and ITIN. The dataset fully covers all 25 FinCEN typology codes and provides detailed statistics and multiple configuration options. Additionally, a separate chain-of-thought reasoning dataset is available for SFT, LLM-as-judge distillation, or CoT data augmentation. The dataset supports multiple languages, including English, Spanish, Vietnamese, etc., and includes detailed citations and licensing information.