From Historical Tabular Image to Knowledge Graphs: A Provenance-Aware Modular Pipeline
Researchers have introduced a novel, modular, and provenance-aware pipeline designed to convert handwritten archival tables into structured Knowledge Graphs (KGs). This approach addresses the limitations of opaque end-to-end AI systems by decomposing the complex multimodal process into three distinct stages: table reconstruction, information extraction, and KG construction. By exposing intermediate representations, the pipeline facilitates human oversight, inspection, and correction, thereby enhancing trust and collaboration between humans and AI. A key innovation is the systematic integration of data provenance at every stage, ensuring that all extracted entities remain traceable to their original visual and textual sources. The method was validated through experiments on real-world archival materials related to military careers, demonstrating the effectiveness of modularity in handling complex historical data. This work significantly advances the transparency and controllability of image-to-KG conversion processes, offering a robust solution for digitizing and structuring rich historical information contained in handwritten documents.
Wire timeline
From Historical Tabular Image to Knowledge Graphs: A Provenance-Aware Modular Pipeline
Researchers have introduced a novel, modular, and provenance-aware pipeline designed to convert handwritten archival tables into structured Knowledge Graphs (KGs). This approach addresses the limitations of opaque end-to-end AI systems by decomposing the complex multimodal process into three distinct stages: table reconstruction, information extraction, and KG construction. By exposing intermediate representations, the pipeline facilitates human oversight, inspection, and correction, thereby enhancing trust and collaboration between humans and AI. A key innovation is the systematic integration of data provenance at every stage, ensuring that all extracted entities remain traceable to their original visual and textual sources. The method was validated through experiments on real-world archival materials related to military careers, demonstrating the effectiveness of modularity in handling complex historical data. This work significantly advances the transparency and controllability of image-to-KG conversion processes, offering a robust solution for digitizing and structuring rich historical information contained in handwritten documents.
cs.AI updates on arXiv.org