Quantitative exploration of large medieval manuscripts data for the codicological research

Abstract

Quantitative exploration is gaining in importance for the analysis of the digitized medieval manuscripts. While codicologists can collect such massive amounts of heterogeneous datasets digitized in high-resolution, they still lack efficient and intuitive means to explore data and answer domain-specific research questions. A new approach is needed to enable codicologists with the quantitative exploration of large of data. This paper presents a concept of a fully integrated system to enable a quantitative exploration of various layout features and their uncertainties over a large collection of medieval manuscripts. It is composed of three main components: data handling that ingests large amounts of data into a data repository, feature extraction that extracts various layout features of the manuscripts and quantitative exploration for visual analysis. In addition to this, we introduce new visualization approaches, i.e. the superimposition plot and the manuscript montage plot in combination with the parallel coordinate plot to explore structural layout features of 170,000 manuscript pages with more than 2.5 million measurements of these layout features. Our approach supports codicologists to see the overall structure of the manuscript at a single glimpse, explore heterogeneous layout features and convey uncertainties. We demonstrate typical use-case scenarios of our collaborators in codicological research where our system has enabled them to answer domain-specific questions for analysis of the medieval manuscripts data for the first time.