A Framework for Online Clustering Based on Evolving Semi-Supervision

Guilherme AlvesMaria Camila N. BarioniElaine R. Faria

The huge amount of currently available data puts considerable constraints on the task of information retrieval. Automatic methods to organizedata, such as clustering, can be used to help with this task allowing timelyaccess. Semi-supervised clustering approaches employ some additional information to guide the clustering performed based on data attributes to a moresuitable data partition. However, this extra information may change over timeimposing a shift in the manner by which data is organized. In order to help copewith this issue, we propose the framework called CABESS (Cluster AdaptationBased on Evolving Semi-Supervision), for online clustering. This frameworkis able to deal with evolving semi-supervision obtained through user binaryfeedbacks. To validate our approach, the experiments were run over hierarchical labeled data considering clustering splits over time. The experimentalresults show the potential of the proposed framework for dealing with evolvingsemi-supervision. Moreover, they also show that our framework is faster thantraditional semi-supervised clustering algorithms using lower standard semi-supervision.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: