Understanding Scam Trends and Rail Paths from Reddit Self-Disclosure Narratives

The study introduces a dataset derived from 21,304 Reddit posts related to scams, collected from 2023 to 2025, to analyze trends and rail paths in scam behavior. It employs a heuristic annotation method to label 1,800 posts with explicit scam chains, evaluated through human annotation, and utilizes topic modeling to examine community support behaviors. This research highlights the multi-rail nature of scams and varying path complexities, providing valuable insights for practitioners focused on synthetic scam chain data simulation and AI-driven scam risk assessments.