Now this one really is a great idea

There’s a couple of computer science/data visualisation geeks who have come up with a really nifty way of graphically representing verbal similarities between a group of essays.

Basically, the gist of the process is that the essays are scanned, and word frequencies are calculated. The most frequently occurring words are then represented by larger and larger circles, so each essay is placed in context of the group, and relations between them can be seen at a glance. If you then drill down into each essay, the bar-code looking ring breaks down the structure of the essay into paragraphs, footnotes, etc.

Beautiful, useful, nerdy: awesome.

It strikes me that this, if implemented well, could be a really useful tool for preliminary research. When researching an essay, you feed a bunch of (hopefully) related articles into it, and you get a birds-eye-view of how they relate to each other, and a simple idea of the major themes of each article. Now it’s not too sophisticated – the major theme is decided by word count – but it’s a start at least. I imagine you could do some further links by examining the references of each article: if two articles both reference another work, they may be tied more closely.

There’s another possibility. One of our college lecturers took us through an outline of Romans, one of the books of the New Testament. A significant part of his analysis came down to verbal parallels between different parts of the letter; that is, certain parts ’sounded’ very similar to others in terms of the greek terms used. Perhaps a tool like this could be useful in getting a preliminary idea of any verbal parallels between different sections of text, which might throw up some useful ideas for further consideration.

Fascinating. I might wait a little and see if they come out with any useful software based on their algorithm… if not, I think I now have a summer project.