A Document Clustering Method Based on One-Dimensional SOM

In this paper a new method is presented and used in clustering document collections. This method is based on the one-dimensional arrays of Self-Organizing Map network (1-D SOM array). The main idea of this method is to obtain the clustering results by calculating the distances between every two adjacent MSPs (the most similar prototype to the input vector ) of well trained 1-D SOM. The process is simple, easy to understand and unnecessary to give the number of clusters beforehand. The experimental results show that this method works well in clustering document collection.