Incremental data mining algorithms process frequent updates to dynamicdatasets efficiently by avoiding redundant computation. Existing incrementalextension to shared nearest neighbor density based clustering (SNND) algorithmcannot handle deletions to dataset and handles insertions only one point at atime. We present an incremental algorithm to overcome both these bottlenecks byefficiently identifying affected parts of clusters while processing updates todataset in batch mode. We show effectiveness of our algorithm by performingexperiments on large synthetic as well as real world datasets. Our algorithm isup to four orders of magnitude faster than SNND and requires up to 60% extramemory than SNND while providing output identical to SNND.
展开▼
