Knowledge Guided Clustering Medieval Polychromy

Expressing the membership of certain events in the historic past as characteristic for a specific cultural era sometimes requires the identification of their accumulated occurrence during a specific period. Many clustering techniques typically applied to the identification of such eras use parameters which cannot be estimated by non-data-scientific users of event data-bases. In the art-historical context of artisan schools applying skills and techniques characteristic of a specific state of cultural achievement, it is necessary to agree on the methodology of such applications defining what is meant by accumulation within a certain epoch. By using a semi-parametric clustering method involving histograms and a parameter readily accessible to specialized art historians, we show that a knowledge guided data scientific approach can surpass the dilemma of parameter-dependence and identify periods in time in which certain artistic techniques have been popular.