Evolução da semissupervisão em detecção online de agrupamentos

Abstract

The huge amount of currently available data puts considerable constraints on the task of information retrieval. Automatic methods to organize data, such as clustering, can be used to help with this task allowing timely access. Semi-supervised clustering approaches employ some additional information to guide the clustering performed based on data attributes to a more suitable data partition. However, this extra information may change over time imposing a shift in the manner by which data is organized. In order to help cope with this issue, this dissertation proposes the framework called CABESS (Cluster Adaptation Based on Evolving Semi-Supervision), for online clustering. This framework is able to deal with evolving semi-supervision obtained through user binary feedbacks. To validate the approach, the experiments were run over seven hierarchical labeled datasets considering clustering splits and merges over time. The experimental results show the potential of the proposed framework for dealing with evolving semi-supervision. Moreover, they also show that the framework is faster than traditional semi-supervised clustering algorithms using lower standard semi-supervision.