Experiments on Incremental Clustering

Clustering of very large document databases is essential to reduce the spacehime complexity of information retrieval. The periodic updating of clusters is required due to the dynamic nature of databases. An algorithm for incremental clustering at discrete times is introduced, Its complexity and cost analysis and an investigation of the expected behavior of the algorithm are provided. Through empirical testing, it is shown that the algorithm is achieving its purpose in terms of being cost effective, generating statistically valid clusters that are compatible with those of reclustering, and providing effective information retrieval