Automated recognition of geographical named entities in titles of Ukiyo-e prints

This paper investigates the application of Natural Language Processing as a means to study the relationship between topography and its visual renderings in early modern Japanese ukiyo-e land-scape prints. We introduce a new dataset with titles of landscape prints that have been annotated by an art historian for any included place-names. The prints are hosted by the digital database of the Art Research Center at the Ritsumeikan University, Kyoto, one of the hubs of Digital Humanities in Japan. By applying, calibrating and assessing a Named Entity Recognition (NER) tool, we argue that 'distant viewing' or macroanalysis of visual datasets can be facilitated, which is needed to assist art historical studies of this rich, complex and diverse research material. Experimental results indicated that the performance of NER can be improved by 30% and reach 50% precision, by using part of the introduced dataset.