Customer Feedback on Cryptomarket: Analysis of Insights Enhanced by Transformer Models

Abstract

This study analyzes customer feedback in the AlphaBay cryptomarket, specifically focusing on the topics discussed and the emotional content expressed in text feedback related to narcotics sales. Transformer- based models, including DeBERTa and GPT-3.5 Turbo, were employed to analyze user comments. The analysis reveals that different categories of drugs shape the nature of user feedback, with significant emphasis on delivery and product quality. Feedback related to refunds is predominantly associated with negative ratings. Furthermore, feedback related to topics such as vendor quality and overall experience was found to be the most emotional. Additionally, higher transaction prices are significantly linked to more emotional comments, indicating that users exhibit heightened emotional responses when engaging in high-value transactions. These findings provide insights into the underlying mechanisms of feedback in illicit online marketplaces, highlighting critical areas such as vendor reliability and product quality that impact user satisfaction. The study suggests that understanding these feedback patterns can help law enforcement and regulatory bodies identify key areas of concern within cryptomarkets, thereby enhancing efforts to combat illegal activities and protect consumers.