The Bitcoin scaling debate goes on and on. The argument is not over whether Bitcoin should scale up: clearly, if it is to become a serious challenger to mainstream payments providers such as Visa and central bank RTGS systems such as Fedwire, it must be able to handle daily transaction volumes in the billions. No, the question is how it should scale up. There are basically two camps: those who follow the original thinking of Bitcoin's creator, Satoshi Nakamoto, that all transactions should be on-chain and democratically validated, and those who think that the way forward is to take most transactions off-chain, leaving only large transactions (perhaps made up of thousands of netted small transactions) on the main blockchain. There is little doubt that the network could scale up to handle Fedwire volumes.
Machine learning and data mining techniques have been used extensively in order to detect credit card frauds. However, most studies consider credit card transactions as isolated events and not as a sequence of transactions. In this article, we model a sequence of credit card transactions from three different perspectives, namely (i) does the sequence contain a Fraud? (ii) Is the sequence obtained by fixing the card-holder or the payment terminal? (iii) Is it a sequence of spent amount or of elapsed time between the current and previous transactions? Combinations of the three binary perspectives give eight sets of sequences from the (training) set of transactions. Each one of these sets is modelled with a Hidden Markov Model (HMM). Each HMM associates a likelihood to a transaction given its sequence of previous transactions. These likelihoods are used as additional features in a Random Forest classifier for fraud detection. This multiple perspectives HMM-based approach enables an automatic feature engineering in order to model the sequential properties of the dataset with respect to the classification task. This strategy allows for a 15% increase in the precision-recall AUC compared to the state of the art feature engineering strategy for credit card fraud detection.
It is standard practice in managing payments to block potentially fraudulent transactions via a set of rules. These rules can be very effective in mitigating fraud risk, and practitioners in the industry are comfortable with this approach. Quite often these rules are able to mitigate the losses from fraudulent transactions without producing a correspondingly high alarm rate.
Venmo's privacy settings have once again been proven to be faulty. A computer science student was able to scrape data from seven million transactions made on the PayPal-owned peer-to-peer payments app, according to TechCrunch. The student, Dan Salmon, said he published the transactions in order to alert users that they should set their payments to private. Venmo's privacy settings have once again been proven to be faulty. 'I am releasing this dataset in order to bring attention to Venmo users that all of this data is publicly available for anyone to grab without even an API key,' Salmon wrote in a post sharing his research.
Zhang, Xinyi (University of California, Santa Barbara) | Tang, Shiliang (University of California, Santa Barbara) | Zhao, Yun (University of California, Santa Barbara) | Wang, Gang (Virginia Polytechnic Institute and State University) | Zheng, Haitao (University of California, Santa Barbara) | Zhao, Ben Y. (University of California, Santa Barbara)
For millions around the globe, digital payment apps such as Venmo are replacing cash as the preferred method of payment between friends and vendors. Apps like Venmo bring a unique blend of convenience and social interactions into financial transactions. In this paper, we study the role of social relationships in the adoption of the Venmo digital payment system. We collect records of all 91 million public transactions conducted on Venmo since its introduction, a social graph connecting most of its 10.5 million users, and analyze the interplay between social relationships and financial transactions. We find that Venmo communities are very densely connected compared to other interaction networks, and are often driven by specific niche applications. We are able to extract both user-to-user and user-to-vendor transaction communities, and show that they exhibit dramatically different structural properties.