MultiClust special issue on discovering, summarizing and using multiple clusterings

Traditionally, clustering has focused on discovering a single grouping of the data. In many applications, however, data is collected for multiple analysis tasks. Several features or measurements provide complex or high dimensional information. In such data, one typically observes several valid groupings, i.e. each data object fits in different roles. In contrast to traditional clustering these multiple clusterings describe alternative aspects that characterize the data in different ways. Traditional single clustering solutions can thus be regarded as special cases of multiple clustering solutions, where only a single set of clusters represents one notion of intra-cluster similarity and inter-cluster dissimilarity. The generality of multiple clustering solutions allows capturing multi-faceted information in more than a single similarity notion, while making it more challenging to detect cluster structures.