I've wanted to do this as well. How does your site rank stories within a cluster? I had always imagined that if I were to do this, I would order them by centrality, where the distance metric is Euclidean distance between the word histograms.

I'm doing graph clustering, where stories are connected by shared terms, so I'm ranking based on the weight of the edges to a story inside the cluster. The terms are weighted by frequency over all the stories in the system with the intention of prioritizing articles contain a lot of specific terms that look like they're related to the story.