Due to the increasing trend of data gathering, data clustering has become an inevitable and highly used analysis technique in many application domains. From the end user’s perspective, the wide variety of available algorithms and their technical parameterization bring major diffculties in the determination of a user-satisfying clustering result. AUGUR aims to overcome this issue in the context of largescale analysis by developing a novel feedback-driven clustering process.
This clustering process contains two main elements: An algorithmic platform and a visual-interactive interface. The algorithmic basis of AUGUR is derived from the concept of ensemble-clustering, which offers robust result with increased quality. In contrast to traditional algorithms, AUGUR substitutes technical parameterization with a compact set of intuitive and user-friendly feedback operations. These operations are applied via a visual interface that displays clustering results and guides the user in choosing appropriate feedback to refine the obtained results. While the algorithmic component of AUGUR runs on stationary high-performance hardware, the visual interface is a lightweight android application that can be used anytime and anywhere.