Semi-supervised Labeling Model Based on Gaussian Mixture in the Context of E-commerce Price Fraud

E-commerce transaction fraud has become an increasingly serious problem in recent years. Mining and analyzing e-commerce transaction data can identify potential transaction frauds, promote fair competition in the market and facilitate regulation. However, there are inevitable problems of missing values and unlabeled data obtained from e-commerce platforms. In this paper, we propose a semi-supervised data labeling method, which interpolates missing values before labeling, solves the problem of removing a large amount of valuable feature data to remove missing values, and then selects the semi-supervised labeling model with the highest accuracy through experiments. The data processed by the above method is helpful to improve the accuracy of the model, which is a practical guidance for e-commerce transaction fraud detection.