Apache Common Math fuzzy kmean clustering does not produce overlapped clusters

[Ameer Albahem]

Hi all,

I am trying to cluster a set of text documents using Apache Common math.
The document might belong to multiple clusters, so i used the Fuzzy kmean
clustering implementation. Things are working fine except that the final
clusters do not contain overlapped documents. My understanding is that the
final clusters should have documents that overlapped across documents. I
tried setting different values for fuzziness parameters, but it did not
work. The final clusters always contain non overlapping documents.

I have a class called "DocumentPoint" that extends the Clusterable
interface, then i am using something like this:

     FuzzyKMeansClusterer<DocumentPoint> clusterer = new
FuzzyKMeansClusterer<DocumentPoint>(k,fuzziness);

     clusterer.cluster(docPoints);

     return clusterer.getClusters();


Help is very much appreciated.


Regards

Ameer