Automatically Dismantling Online Dating Fraud (1905.12593v2)

Abstract: Online romance scams are a prevalent form of mass-marketing fraud in the West, and yet few studies have addressed the technical or data-driven responses to this problem. In this type of scam, fraudsters craft fake profiles and manually interact with their victims. Because of the characteristics of this type of fraud and of how dating sites operate, traditional detection methods (e.g., those used in spam filtering) are ineffective. In this paper, we present the results of a multi-pronged investigation into the archetype of online dating profiles used in this form of fraud, including their use of demographics, profile descriptions, and images, shedding light on both the strategies deployed by scammers to appeal to victims and the traits of victims themselves. Further, in response to the severe financial and psychological harm caused by dating fraud, we develop a system to detect romance scammers on online dating platforms. Our work presents the first system for automatically detecting this fraud. Our aim is to provide an early detection system to stop romance scammers as they create fraudulent profiles or before they engage with potential victims. Previous research has indicated that the victims of romance scams score highly on scales for idealized romantic beliefs. We combine a range of structured, unstructured, and deep-learned features that capture these beliefs. No prior work has fully analyzed whether these notions of romance introduce traits that could be leveraged to build a detection system. Our ensemble machine-learning approach is robust to the omission of profile details and performs at high accuracy (97\%). The system enables development of automated tools for dating site providers and individual users.

• The paper presents an ensemble machine-learning approach that achieves 97% accuracy in identifying romance scams.
• It integrates multi-modal feature extraction from demographics, images, and text to uncover scammers’ manipulations in dating profiles.
• The publicly available tool empowers platforms to preemptively dismantle fraudulent profiles and mitigate significant victim harm.

Overview of "Automatically Dismantling Online Dating Fraud"

The paper under discussion presents a comprehensive paper into the detection of online romance scams, a predominant form of mass-marketing fraud that exploits online dating platforms. This research is particularly critical considering the substantial financial and psychological toll these scams inflict upon victims. The authors introduce an ensemble machine-learning based system designed to identify romantic scammers by analyzing dating profiles with a remarkable accuracy of 97%.

Key Contributions

The significant contributions of this paper can be divided into several aspects:

1. Data Collection and Analysis: The researchers gathered an extensive dataset from \url{datingnmore.com} and \url{scamdigger.com}, which included both genuine and scam profiles. The demographic analysis revealed strategic choices made by scammers regarding profile presentation, including manipulated ethnic, occupational, and locational information aimed at appealing to specific victim traits.
2. Profile Feature Extraction: The authors utilize structured and unstructured data from user profiles. They categorize features into demographics, images, and textual descriptions, leveraging image analysis and natural language processing techniques to extract relevant characteristics. This multi-modal feature extraction is pivotal for identifying unique patterns associated with scam profiles.
3. Ensemble Classification Approach: A machine-learning ensemble method combines predictions from multiple classifiers tailored to different profile features—demographics, images, and textual content. The ensemble approach, specifically a weighted voting mechanism, achieves a high level of precision and recall, outperforming individual classifiers and showcasing robustness even when certain profile details are missing.
4. Publicly Available Tool: To facilitate further research and application, the tool developed is made publicly accessible, encouraging replication and enhancement by other researchers and practitioners.

Numerical Results and Methodology

The ensemble classifier's performance was validated using a robust experimental setup involving training, validation, and testing datasets. The classifier's high F1-score of 0.945 indicates a substantial capability in discriminating between real and scam profiles. The authors also conducted an insightful feature analysis, highlighting the discriminative power of certain demographics and linguistic features commonly associated with scam profiles.

Implications and Future Directions

From a practical perspective, this system offers a preventative mechanism that dating platforms can adopt, potentially reducing scam incidents by identifying fraudulent profiles early. The approach also underscores the importance of using diverse data sources and analytical methods, particularly in domains with high social and financial stakes like online dating.

From a theoretical standpoint, this paper extends the boundaries of automated fraud detection, emphasizing the efficacy of ensemble learning models in real-world, heterogeneous data contexts.

Looking forward, potential developments could include adapting the system to respond to evolving scam tactics, such as profile cloning, and integrating behavior-based analysis beyond static profile attributes. This would further enhance the model's adaptability and resilience against sophisticated scammer evasion strategies.

In summary, this paper provides a substantive advancement in the automated detection of romance scams, emphasizing the synergies of data-driven methodologies and collaborative research to combat online fraud effectively.