Applying Graph Neural Networks to ERP Systems with a Focus on Business Centric Development

Software for enterprise resource planning (ERP) has grown in popularity as a means to streamline corporate processes that are critical for survival in the modern, globalized marketplace. In order to cater to the specific needs of their global customer base in terms of both price and availability, suppliers in the textile industry are under increasing pressure to enhance their manufacturing management practices. This task is becoming more dependent on the careful selection and usage of information technology. Preprocessing, feature extraction, and training are the three phases that make up this method. Time difference relevance and K-L distance are components of preprocessing. The feature selection approach employs Kernel Principal Component Analysis for the factor extraction phase. An GNN was used for the model’s training. The proposed technique outperforms CNN and SVM with an average accuracy of $\mathbf{9 2. 1 6}$ percent.