Constructing and Mapping Fuzzy Thematic Clusters to Higher Ranks in a Taxonomy

We present a novel methodology for mapping a system such as a researchdepartment to a related taxonomy in a thematically consistent way. Thecomponents of the structure are supplied with fuzzy membership profiles overthe taxonomy. Our method generalizes the profiles in two steps: first, by fuzzyclustering, and then by mapping the clusters to higher ranks of the taxonomy. Tobe specific, we concentrate on the Computer Sciences area represented by the taxonomyof ACM Computing Classification System (ACM-CCS). We build fuzzyclusters of the taxonomy leaves according to the similarity between individualprofiles by using a novel, additive spectral, fuzzy clustering method that, in contrastto other methods, involves a number of model-based stopping conditions.The clusters are not necessarily consistent with the taxonomy. This is formalizedby a novel method for parsimoniously elevating them to higher ranks of the taxonomyusing an original recursive algorithm for minimizing a penalty functionthat involves "head subjects" on the higher ranks of the taxonomy along withtheir "gaps" and "offshoots". An example is given illustrating the method appliedto real-world data.