Automatic medieval charters structure detection : A Bi-LSTM linear segmentation approach

This paper presents a model aiming to automatically detect sections in medieval Latin charters. These legal sources are some of the most important for studies as they reflect economic and social dynamics well institutional writing practices. An automatic linear segmentation can greatly facilitate charter indexation speed up recovering evidence support historical hypothesis by means granular inquiries on these raw, rarely structured sources. Our is based Bi-LSTM approach using final CRF-layer was trained large, annotated collection charters (4,700 documents) coming from Lombard monasteries: CDLM corpus (11th-12th centuries). The evaluation shows high performance test-set an external consisting Montecassino abbey (10th-12th We describe architecture model, main problems related treatment formulaic discourse, we discuss implications results terms record-keeping practices High Middle Ages.