This chapter establishes a multi-modal analytical framework that synthesizes external litigation information, traditional financial reporting, and internal management data, employing regulatory enforcement actions that demonstrate significant correlation with financial fraud as proxy variables to construct an imbalanced large-scale dataset.This chapter explores financial fraud detection models by integrating resampling techniques with machine learning methods, considering scenarios with and without feature selection, to identify optimal combinations of resampling methods and classification models. Furthermore, the role of litigation factors in financial fraud detection within immature legal environments is examined, providing a new research perspective for financial fraud detection studies.