The Comparison of SOM and K-means for Text Clustering

摘要

SOM and k-means are two classical methods for text clustering. In this paper some experiments have been done to compare their performances. The sample data used is 420 articles which come from different topics. K-means method is simple and easy to implement; the structure of SOM is relatively complex, but the clustering results are more visual and easy to comprehend. The comparison results also show that k-means is sensitive to initiative distribution, whereas the overall clustering performance of SOM is better than that of k-means, and it also performs well for detection of noisy documents and topology preservation, thus make it more suitable for some applications such as navigation of document collection, multi-document summarization and etc. whereas the clustering results of SOM is sensitive to output layer topology.