Opening the black box of bibliometric mapping algorithms: Which algorithm should I choose?

The task of bibliometric mapping involves the usage of a data model together with an algorithm to analyze the data. In the last years, we learned more about the behavior of data models like co-citation or bibliographic coupling and their usefulness for certain tasks. From these studies, however, we learn relatively little about the impact of the choice of the algorithm. Usually, bibliometric mapping algorithms are imported from other disciplines, which means that they have not been developed for this particular purpose. We provide here a framework to analyze if the different assumptions on the data that these algorithms make correspond with the properties of topics they are trying to reconstruct.