Fraudulent financial reporting continues to undermine market integrity and threaten stakeholder interests globally. Conventional analytical methodologies demonstrate limited effectiveness in recognizing intricate temporal relationships and nuanced irregularities characteristic of accounting manipulation. This research introduces an enhanced recurrent neural architecture incorporating multi-head attention layers with bidirectional LSTM components, specifically engineered for identifying accounting anomalies. Our architectural design combines attention-weighted feature selection with dual-directional temporal processing to simultaneously extract prolonged sequential patterns and pivotal indicators spanning multiple reporting cycles. Through rigorous evaluation using a dataset of 2,500 corporate entities across 12 reporting cycles, our methodology achieves 93.87% classification accuracy, 92.41% positive predictive value, 91.56% sensitivity, and 91.98% harmonic mean of precision-recall. These metrics surpass conventional statistical methods and standard deep architectures by substantial margins. The attention weight distributions offer transparent interpretability, revealing which specific accounting ratios and temporal windows contribute decisively to fraud predictions, thereby enhancing practical deployment feasibility in professional audit environments.

This study proposes a method for automatic audit report generation and compliance analysis based on the BERT model, aiming to solve the problems of low efficiency, high labor costs, and insufficient compliance detection accuracy in traditional audit work. By combining the powerful semantic understanding ability of the pre-trained language model, this study achieves high-quality generation of audit reports and uses classification tasks to accurately identify potential risk points in the text. The experiment uses multiple public datasets to evaluate the performance of BERT in text generation quality (BLEU Score) and compliance detection (Accuracy, F1-score). Compared with models such as GPT, T5, and LSTM, BERT performs best in all indicators. At the same time, through robustness testing, the performance of the model in the face of text noise or tampered data is analyzed. The results show that although noise has a certain impact on model performance, BERT's overall performance is still better than that of other comparison models. This study also explores the impact of different fine-tuning strategies on model performance and analyzes its adaptability and limitations in the audit field. The research findings demonstrate that the BERT-based methodology not only effectively enhances the quality of the automated generation of audit reports but also substantially improves the accuracy of compliance analysis. This advancement offers novel insights and serves as a guiding framework for the development of intelligent audit technology.

This paper proposes an anomaly detection algorithm based on large language models guided by knowledge graphs. The method combines the deep semantic understanding capability of large language models with the structural modeling advantages of knowledge graphs, forming a semantic-structural co-driven fraud detection framework. Specifically, the model first encodes transaction texts using a pretrained language model to generate semantic representations. Meanwhile, a financial knowledge graph is constructed based on entities such as transaction accounts, devices, and IP addresses, from which structural embeddings are obtained via graph neural networks. Next, a gated fusion mechanism is introduced to adaptively integrate the semantic and structural vectors. The fused representation is then fed into a classifier to produce fraud prediction results. In multiple comparative experiments, the proposed model outperforms single-modality approaches across metrics including accuracy, precision, recall, and F1-score, validating its effectiveness in complex financial scenarios. In addition, this study conducts multi-task adaptation experiments and anomaly ratio variation tests to evaluate the model's stability and robustness. The results show that the proposed method maintains strong performance across different environments, demonstrating its reliability and practical value.

This study proposes a graph neural network (GNN)- based method for fraud detection in financial transaction networks. By deeply modeling the transaction graph structure and the relationships between nodes, the method enhances the ability to identify complex fraudulent behaviors. First, financial transaction data is transformed into a graph structure, where transaction accounts are represented as nodes and the flow of funds as edges. Node features such as transaction amount and timestamp are used for node representation learning. Next, graph neural network architectures, such as Graph Convolutional Networks (GCN) and Graph Attention Networks (GAT), are employed to uncover the latent associations and interactions between nodes. This helps identify anomalous fraudulent transactions. To improve the model's generalization and robustness, the study also introduces contrastive learning strategies and imbalance handling techniques. A weighted loss function is used to optimize the model's performance on minority fraud samples. Experimental results show that the proposed method outperforms traditional machine learning models and other deep learning approaches across various evaluation metrics on publicly available datasets. Notably, it achieves a better balance between precision and recall. This method effectively combines graph structural information with deep learning techniques, providing a novel solution for intelligent financial risk control.