Speaker: Dr Benoit Seguin
With the increased availability of digitized documents in IIIF format, researchers can now easily compare materials from different institutions right in their web browser. However, this approach is still limited only to viewing documents one might know of, or have discovered on the institution's website directly. Because the capabilities of machine learning for document analysis, text-recognition, visual similarity and text understanding have improved drastically in the last few years, it has become possible to semi-automatically index complete corpora textually and visually. This presentation will report on the early stages of a project trying to apply these techniques to architectural treatises, but will try to generalise to the potential applications to other corpora.