Experiment with combining summarization and conversation mode

The idea is quite simple: to generate creative results from multiple summarized texts. To do this, we need an algorithm that can effectively find related texts among the summaries. The reason for this is to enable a conversation mode, where the summarizers can communicate with each other using only phrases from the summaries. I am curious to see what kind of conversations they can produce.