A Fraudster in a Haystack: Crafting a Classifier for Non-delivery Fraud Prediction at Online Auction Sites

Almendra, Vinicius; Enachescu, Denis

doi:10.1109/synasc.2012.21

Cited by 2 publications

(1 citation statement)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Various data mining techniques have been applied in FFD, such as decision trees (Kirkos et al, 2007), (Kotsiantis et al, 2006), neural networks (Cerullo and Cerullo, 1999), (Dorronsoro et al, 1997), (Fanning and Cogger, 1998), , (Viaene et al, 2005), Naive Bayes method (Bermudez et al, 2008), (Viaene et al, 2004), and logistic regression models (Bell and Carcello, 2000), (OwusuAnsah et al, 2002), (Spathis, 2002), (Yuan et al, 2008). Almendra and Enachescu (2012) discovered a set of features that can be extracted from the public information of online auction sites. used a set of characteristics for fraud detection in e-markets that was recently found.…”

Section: Literature Reviewmentioning

confidence: 99%

Machine Learning Based Prediction and Prevention of Malicious Inventory Occupied Orders

Yang

Cheng

et al. 2014

International Journal of Mobile Computing and Multimedia Communications

View full text Add to dashboard Cite

MIOOs are orders created temporarily for the purpose of occupying the inventories of sellers. MIOOs disrupt normal business activities and harm both sellers and consumers. This study aims to determine the best practice and model of the technical solutions that can effectively and systematically limit malicious inventory occupied orders (MIOOs), using the methods of analytical mining and case studies. This work contains three contributions. Firstly, this work solves MIOOs problem by using machine learning technology. The result of the study indicates that 93% of MIOOs from the sample data are actually predictable and preventable. Secondly, this work presents a methodology of solving MIOOs problem which can be applied by other companies. The methodology in this paper consists of four major steps, namely doing statistics concerning MIOOs, using logistic regression algorithm to train a mode, optimizing the model, and applying the model. Finally, this work finds unique features of MIOOs, and they can help better understanding the behind logic of MIOO producers.

show abstract