Information extraction from receipts involves transforming unstructured data into structured formats through a process that includes Optical Character Recognition (OCR) and tagging. The OCR phase extracts textual data from images, while the tagging phase assigns semantic labels to these text fragments, using visual layout information inherent in receipts to enhance accuracy. Traditional methods like template-based and NLP-based approaches have limitations in handling varying receipt formats and complex layouts. To address these challenges, Graph Convolutional Networks (GCNs) are employed, leveraging graph structures to model relationships between text elements. GCNs use nodes to represent words and edges for their connections, enabling the classification of text elements based on patterns recognized during training. This technique is particularly useful for visually rich documents where spatial arrangements convey critical information. The pipeline for using GCNs in receipt information extraction includes steps such as graph modeling, feature calculation, and semi-supervised learning, facilitating the accurate tagging of receipt components like company names, dates, and amounts.