Is there anyway to get more fine grained clustering of sentences? When I’m clustering semi structured/regular text (e.g. product descriptions that have groups with minor variations) I’m getting large clusters, but would like to break down further.
The granularity feature is currently not implemented, but thanks for bringing it up. Since the algorithm does soft clustering, one way to get a hold of finer granularity clusters is to drop the large clusters. The larger clusters are usually encapsulation of broader topics and the finer grained topics within those broader clusters usually appear in smaller groups. If you can send me example input and output I could give you more ideas as well.