Large-scale genealogical information extraction from handwritten Quebec parish records

This paper presents a complete workflow designed for extracting information from Quebec handwritten parish registers. The acts in these documents contain individual and family highly valuable genetic, demographic social studies of the population. From an image records, our is able to identify extract personal information. divided into successive steps: page classification, text line detection, recognition, named entity recognition act detection classification. For all steps, different machine learning models are compared. Once extracted, validation rules by experts then applied standardize extracted ensure its consistency with type (birth, marriage, death). step reject records that considered invalid or merged. full has been used process over two million pages registers 19-20th centuries. On sample comprising 65% registers, 3.2 were recognized. Verification birth death this shows 74% them valid. These will be integrated BALSAC database linked together recreate genealogical relations at large scale.