• Features:

This Java program generates a sequence of web pages to illustrate clustering results. Here is an example of the clustering results of about 114,000 NSF award abstracts.

• Input: document (and word) clusters produced by "spkmeans".
• Output: Web pages that show the structure of the clusters.

• Code:

The code is released under the GNU Public License (GPL).
 

• To compile: javac ClusterBrowser.java

It will generate 3 '.class' files: ClusterBrowser.class, Document.class, and SparseMatrix.class.
 
• To run: java -mx128m ClusterBrowser /dir/classic3_tfn 3 file:// [classic3_tfn_wordtoclus.3]
• All the files including the word-document matrix files and the clustering results should be in the same 'dir'.
• '-mx128m' is used to reserve 128MB main memory. Of course it's up to how many documents and clusters you have.
• 'file://' is the protocol you'd like to access the web pages to be generated by ClusterBrowser.
  
• '3' is the cluster number.
• 'classic3_tfn_wordtoclus.3' is the word cluster file name. Given only when you want both document and word clusters to be illustrated.

 
You are welcome to use the code under the terms of the licence for research or commercial purposes, however please acknowledge its use with a citation:

Dhillon, I. S. and Fan, J. and Guan, Y.,   "Efficient Clustering of Very Large Document Collections", 2000

Here is a BiBTeX entry:

@ARTICLE{dhillon:fan:guan00,
      AUTHOR = {Dhillon, I. S. and Fan, J. and Guan, Y.},
      TITLE = {Efficient Clustering of Very Large Document Collections},
      BOOK = {Data Mining for Scientific and Engineering Applications},
      YEAR = {2001}
}

Bug reports and comments are appreciated.