A TreeCloud is a visualization technique that uses a tree structure to display words used in close proximity of each other. I thought the technique would be useful if applied to tweets from an event, given the temporal nature of posts. I used TwatterKeeper to export tweets containing the #ozchi hashtag for the first two days of the OzChi conference. The TreeCloud visualization was able to group together tweets about the keynotes (John Seely Brown, Elizabeth Churchill and Jacob Buur) and other themes such as vote for the student posters, get the OzChi iPhone app, travelling to OzChi in Brisbane and me (@aneesha) tweeting about analysing Ozchi paper abstracts over 5 years. I have included an annotated version of the TreeCloud for OzChi 2010 tweets below.

Software to create TreeClouds is available from http://www.lirmm.fr/~gambette/treecloud/. The code is written in Python. An online tool to create TreeClouds is also available.
Thanks for your interest in TreeCloud! If you want you can also try to feed TreeCloud with you corpus of the previous post. Also, if you use TreeCloud in Perl + SplitsTree (if you find it difficult to install the program, just send me one text file with all abstracts and I’ll do it for you) instead of the online version, you’ll be able to color the tree cloud chronologically, which is nice for a conference as you somtimes see how the topics evolve… But maybe that was already your plan for part 2 of this post.
Sorry, I meant “TreeCloud in Python + SplitsTree”, there is no Perl version yet.