Multi-Perspective Hotel Operation Process Anomaly Prediction Method Based on Graph Transformer and Autoencoder

Yidan Ma^1*Yue Wu²

• ¹Vocational College Of Defense Technology, Luan, China
• ²Tiangong University, Tianjin, China

Due to the influence of various factors, abnormal situations are inevitable in hotel operation processes. Predicting upcoming anomalies in advance can ensure the reasonable operation of the process. Current advanced anomaly prediction methods use deep learning to encode control flow and data flow information for predicting anomalies in attributes like activity and time. However, this approach struggles to represent the behavioral relationships between process activities and the interaction between control flow and data flow is not specific. To address this issue, this paper proposes a business process anomaly prediction method based on Multi-perspective Graph Transformer and Auto Encoder (MLGTAE). This method first uses Petri nets to capture process behaviors and combines time, resource, and other data attributes to construct multi-perspective trace graphs. Finally, the attention mechanism is used to achieve deep interaction between behavior and data, and the decoder performs reconstruction to predict anomalies. The method was verified through multiple real datasets, and the results show that the proposed method outperforms the comparison methods in anomaly prediction at both the activity level and the data attribute level.