Nonparametric statistical structuring of knowledge systems using binary feature matches

Application of an Automatic Data Alignment & Structuring System for Intercultural Consumer Segmentation Analysis

“…dtu.dk, Mørup et al, 2010;Mørup and Schmidt, 2012;Schmidt and Mørup, 2013;Schmidt et al, 2012). The principle of the stochastic block-model is to partition a set of objects into clusters whose members are stochastically equivalent and possess similar relations to other clusters to be extracted in the partitioning process.…”

“…From a technical aspect, our hypothesis is that the nonparametric Bayesian relational modelling is a powerful and suitable approach (Schmidt and Mørup, 2013;Mørup and Schmidt, 2012) for extracting and aligning consumer segments existing across a multiplicity of markets. The model jointly partitions consumers in the respective markets into segments representing either universal value patterns across cultures or culturally specific value patterns.…”

“…The establishment of semantic relations across boundaries of e.g. organizational-, linguistic-, or societal communities (Fensel et al, 2004) has been a central topic of the ontology matching discipline (Euzenat and Svaiko 2007) and substantial numbers of ontology matching frameworks have been introduced in recent years (Berlin and Motro, 2002;Cheng et al, 2008;Doan et al, 2004;Euzenat, 1994;Stumme and Mädche, 2001;Wang et al, 2004;Mørup et al, 2014). However, a fully automatic ontology alignment has remained a highly challenging task since it typically involves similarity computations between objects belonging to different knowledge systems.…”

“…As indicated in previous works (Isaac et al, 2007;Pirrò and Seco, 2008;Pirrò and Euzenat, 2010;Ngo et al, 2013;Cross et al, 2013;Glückstad et al, 2014;, similarity measures influence on the performance of ontology alignment tasks. Mørup et al (2014) addresses this ambiguity by introducing a new framework applying the nonparametric Bayesian relational modeling to statistically analyze feature matches of all combinations of objects belonging to different knowledge systems, i.e., to align them while jointly partitioning them into clusters. This new way of analyzing "relatedness" between objects belonging to different knowledge systems engenders an additional ability to automatically align three or more ontologies simultaneously (Mørup et al, 2014).…”

“…Whereas this ability to automatically align and structure multiple datasets contributes substantially to the knowledge engineering and ontology development discipline (e.g. Glückstad et al, 2014;Mørup et al 2014), such ability could also be useful in other research disciplines which requires the analysis of "relatedness" of objects across multiple datasets. This position paper introduces our ambitious international research project proposal which sheds light on the potential contributions of such automatic data alignment and structuring framework to other research disciplines, namely, the cross-cultural consumer segment analysis.…”