Comparing Machine Learning Approaches for Table Recognition in Historical Register Books
This work addresses the challenge of extracting information from historical documents for archivists and researchers, but it is incremental as it compares existing methods on a new dataset.
The paper tackled the problem of table recognition in historical handwritten register books by comparing Conditional Random Fields and Graph Convolutional Networks for row and column detection, achieving an F1 score of 89% on death records from the Archive of the Diocese of Passau.
We present in this paper experiments on Table Recognition in hand-written registry books. We first explain how the problem of row and column detection is modeled, and then compare two Machine Learning approaches (Conditional Random Field and Graph Convolutional Network) for detecting these table elements. Evaluation was conducted on death records provided by the Archive of the Diocese of Passau. Both methods show similar results, a 89 F1 score, a quality which allows for Information Extraction. Software and dataset are open source/data.