Optimised Workflow for Knowledge Graph Construction

This technical report presents a proposed workflow for constructing a knowledge graph from archival records related to the 'Wiedergutmachung' process in Germany. The focus is on transforming unstructured historical documents into structured data points, facilitating access and insights into compensation and restitution efforts following World War II. The workflow includes converting document images to machine-readable formats using Optical Character Recognition (OCR), ontology design, information extraction, and entity linking. The report outlines the unique challenges posed by historical documents and discusses the integration of traditional methods with advanced transformer-based technologies. It emphasizes the role of knowledge graphs in connecting extracted information on the Semantic Web, enabling researchers to uncover relationships and patterns within the data. The report also highlights the significance of increasing accessibility to archival data for public participation and understanding of historical injustices, while detailing the intricacies of the digitalisation pipeline and its potential applications in Digital Humanities research.