As the digital landscape evolves, safeguarding customer activities becomes increasingly crucial. This paper addresses the pressing challenge of identifying fraud in customer transactions by employing advanced ML techniques and developing an API for enhanced data visualization. The introduction contextualizes the importance of fraud detection amidst rising digital interactions and associated security vulnerabilities. It highlights the limitations of conventional methods and underscores the pivotal role of ML in enhancing the speed and accuracy of fraud detection. The literature review delves into prior research on fraud detection, ML algorithms, and API development for similar purposes. Identifying gaps in existing knowledge sets the stage for the original contributions sought in this paper. The research objectives delineate precise study goals aimed at selecting and employing ML methods tailored for fraud detection. The methodology section discusses the dataset used, the preparatory steps taken, and the rationale behind the selection of specific algorithms. The emphasis is placed on transparency and reproducibility to fortify the experimental framework. The API implementation segment elucidates the challenges encountered in creating and deploying an API for scam detection. It delineates the technologies employed, their development process, and the integration of ML models into the API. The results section presents performance metrics for the ML models, including accuracy, precision, recall, and other relevant indicators. Demonstrating the API's efficacy in presenting fraud detection data in a way that is both comprehensible and practically usable is paramount. The discussion contextualizes the findings within existing research, emphasizing the unique contributions of the paper to the field. Limitations of the approach are acknowledged, along with suggestions for further investigation. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.