An Incremental Clustering Approach Based On Three-Way Decisions

Most of the clustering algorithms reported assume a data set does not always change. However, it is often observed that the analyzed data set changes over time. To combat changes, we introduce a new incremental soft clustering approach based on three-way decisions theory in this paper. Firstly, an initial clustering algorithm is proposed by using representative points. Secondly, to eliminate the influence of the processing order on final incremental clustering results, the incremental data is pre-clustered the same way. To quickly search similar areas for incremental data, a searching tree based on the representative points is constructed, and the strategies of searching and updating are presented. Finally, the three-way decisions strategy is used for incremental clustering. The results of the example analysis show the approach is effective to incremental clustering.