An ontology-based hierarchical clustering approach for decision support in mass customization environments

The shift from mass production to mass customization has led many manufacturers to diversify their product portfolio by making their products highly configurable. This diversification often results in higher production costs. Through standardization this costs can be reduced. Clustering of sold product configurations can help identify configurations that can be standardized without dissatisfying the customers. Many clustering algorithms are developed for different applications, but often the results are not self-describing. In this paper an ontology-based hierarchical clustering approach with a linkage method particularly designed to return reasonable clusters is proposed. An experimental evaluation with historic sales data supplied by a truck manufacturer shows that the approach is able to find meaningful insights in mass customization contexts. The introduced linkage method shows advantages to the single linkage method in semantic clustering.