One Clustering Process Fits All- A Visually Guided Ensemble Approach

Looking back on the past decade of research on clustering algorithms, we witness two ma-jor and apparent trends: 1) The already vast amount of existing clustering algorithms, is con-tinuously broadened and 2) clustering algorithms in general, are becoming more and more adapted to specific application domains with very par-ticular assumptions. As a result, algorithms have grown complicated and/or very scenario-dependent, which made clustering a hardly ac-cessible domain for non-expert users. This is an especially critical development, since, due to increasing data gathering, the need for analysis techniques like clustering emerges in many ap-plication domains. In this paper, we oppose the current focus on specialization, by proposing our vision of a usable, guided and universally appli-cable clustering process. In detail, we are go-ing to describe our already conducted work and present our future research directions.