The Behavioral Sign of Account Theft: Realizing Online Payment Fraud Alert
Authors: Cheng WANG
IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We examine our method on a real-world B2C transaction dataset from a commercial bank. Experimental results show that the ex-ante detection method can prevent more than 80% of the fraudulent transactions before they actually occur. |
| Researcher Affiliation | Academia | 1Department of Computer Science, Tongji University, Shanghai, China 2Key Laboratory of Embedded System and Service Computing, Ministry of Education, Shanghai, China 3Shanghai Institute of Intelligent Science and Technology, Tongji University cwang@tongji.edu.cn |
| Pseudocode | No | No explicit pseudocode or clearly labeled algorithm blocks were found. |
| Open Source Code | No | No explicit statement about open-sourcing code or providing a code repository link was found. |
| Open Datasets | No | We collect 3.5 million real-world B2C transaction records from a commercial bank. All these records have been labelled as legitimate/fraudulent manually, and the whole data is encrypted and desensitized for security and privacy issues. |
| Dataset Splits | No | The paper mentions training and testing data splits but does not explicitly specify a validation set split or methodology (e.g., percentages, counts, or cross-validation setup). |
| Hardware Specification | No | No specific hardware details (e.g., GPU/CPU models, memory specifications) used for running experiments were mentioned. |
| Software Dependencies | No | The paper mentions machine learning models and some architectural details for DNN, but does not provide specific version numbers for software libraries or dependencies used (e.g., 'XGBoost 1.x', 'Python 3.x'). |
| Experiment Setup | Yes | We set the window size w = 2 and repeat this process 3 times. Table 3 shows the average precision, recall and F1-score on testing data. Note that we set FPR = 0.001, FPR = 0.0005 and FPR = 0.0001, respectively. In the DNN model, the sigmoid function is used as activation function and there are 3 hidden layers in addition to the input and output layers; the neurons at the three hidden layers are 20, 30 and 20, respectively. In our experiment, we increase the class ratio CR from 10 to 70, and set the window size w = 2. In the ex-ante risk prediction module, we set w = 2. When RT = 0.75: Our real-time fraud prevention and detection integrated system can detect 94.46% of fraudulent transactions with less than 0.09% of legitimate transactions interrupted. |