I know as Lingo produces flat clusters.
But, I am seeing subclusters idioms in some comments which makes me feel confused.
Do Lingo also provides subclusters for clusters or they are for other algorithms implemented in the framework?
Other algorithms can produce sub-clusters -- k-means, for example. Our
commercial algorithm Lingo3G can do it too. It would be possible to do
it in Lingo as well if you applied it recursively to top-level
clusters but I think the number of docs per-cluster would be
insufficient to produce something of value.
I use kmeans, with 4000 documents.
The first clusters contains 1989. But I could not get the sub-cluster. Are the number of documents too less?
I'm using the carrot server. I passed in the snippet each 5 contextual sentences.