I was recently pointed to @miljoshi‘s blog and a post on twitter word clouds (using Wordle, of course!). My twitter background was made using Wordle from a sampling of text from my blog. Tweetstats offers the ability to create a Wordle cloud automatically from your tweets, which is fairly cool. Mine is below. It’s dominated by twitpic, since I frequently use it for posting pictures.
Word cloud for my twitter stream.
Update: Here’s my wordle after removing some words that don’t reflect the content of my tweets as much (e.g. good, great, new, old, etc.). Good idea, Melinda!
Updated word cloud for my twitter stream.
That’s pretty cool. I ran mine, and i have a giant @ealdent and @donnadoodle running through mine, since apparently I tend to tweet responses to you guys more often than new tweets :)
Aha, I found where you can automatically exclude @’s. Also cool, you can exclude certain words, so I excluded twitpic from mine so that it wasn’t dwarfing everything else. Wow, Wordle is awesome!
I wonder why tweetstats strips apostrophes. It makes for some weird results, like “youre” in yours, and “husbands” in mine.
I’m guessing the stripping of apostrophes wasn’t thought through. Wordle supports them just fine. But it’s one of those black arts in text processing. Do you remove all punctuation? Which punctuation do you keep? Do dashes mean two separate words or one? What about underscores and slashes? Are commas important?
You just kinda make decisions, maybe because it worked well in the past, interferes with PHP escaping strings, or whatever. Hopefully it would be something you experiment with and see what works better. Like recently I was parsing a lot of text and noticed that words like “dont” were being parsed incorrectly (I think as adjectives or something). By inserting apostrophes, they were then correctly parsed as auxiliary verb + negation.
[...] I wanted my first post on this new blog to be a Word Cloud that represents me…so I found a fellow blogger who shared how to turn your Tweets into a cloud using TweetStats. I had never used TweetStats [...]