Digital libraries have played a very important role in preserving the cultural heritage be it medieval manuscripts, audio material or paintings. Currently, standard manual techniques exist to analyze medieval manuscripts. Analysis performed in this way is very tedious and labor intensive which restricts the number of medieval manuscripts that can be analyzed. Digital methods, tools and techniques are reaching its technological limits due to lack of visual interfaces. This paper presents a visualization concept called CodiVis to facilitate explorations of co-relations in the abstract feature space of large sets of digitized medieval manuscripts. As a starting point, CodiVis uses the medieval manuscripts digitized within the scope of the project "Virtual Scriptorium St. Matthias". It combines two visualization techniques in order to overcome the shortcomings of the single visualization technique. In the first technique, manuscripts are clustered according to the bibliographic metadata and represented in radial tree. This gives a quick overview of the whole dataset. In the second technique, bibliographic metadata is further linked to the macro- and microstructural layout features in the parallel coordinate view. Interactive changes in radial tree are automatically reflected in the parallel coordinate view. The proposed visualization concept shows the potential of analysis by enabling quick exploration of big humanities data. Furthermore, the evaluation tests and feedback from humanities scholars and other users showed that CodiVis is capable of identifying co-relations in arts and humanities.
Swati Chandna, Danah Tonne, Rainer Stotzka, Hannah Busch, Philipp Vanscheidt, Celia Krause, "An effective visualization technique for determining co-relations in high-dimensional medieval manuscripts data" in Proc. IS&T Int’l. Symp. on Electronic Imaging: Visualization and Data Analysis, 2016, https://doi.org/10.2352/ISSN.2470-1173.2016.1.VDA-488