Twitter Sentiment Analysis
Check this out. It's a Summize sentiment analysis of Twitter mentions of the word "Drupal".
It shows that at this particular moment in time, sentiment for Drupal is "swell". Not quite as good as "great", but better than "wretched".
Things are not so swell for Rails this morning because of the rumors about Twitter moving off it. Bummer.
These graphics come from the Labs section of the Summize site. I discovered Summize for their Twitter search capability, which I use it to subscribe to an RSS feed of tweets on several projects as a kind of virtual smoke detector.
I wish the Summize sentiment analysis provided trend charts so you could track sentiment over time. Sentimime does this, but not on an on demand basis. I'd also like to get an embeddable widget that I could put up on the Drupal economic indicators page, and the guys at Summize say they are working on it. For now, I'll just come back and check the labs site every now and then.
It's amazing how many great tools we have these days for spotting and tracking trends. It's a great time to be a data hound!
your data hound skills are terrific love this stuff. i wish summarize actually gave some details behind their colored boxes. they do nothing on the results page to help me understand why we are "swell". or i am missing the evidence.
Thanks, moshe. I don't know the exact process that they use to judge sentiments. But when I looked at the results, I saw more tweets that were generally positive than those that were negative, suggesting that the "swell" rating was on target.
I love your Blogposts, Jeff. Looking forward to seeing you at Szeged and telling you personally... You bring in a fresh perspective.
This tool looks good. A lot of green - good sentiments? Wonder what the location of the dots means. The more on top, the more important?
Thanks! I hope to make it to Szeged, but it's 50-50 at this point.
I think the grid is chronological from left to right, top to bottom. White boxes show no sentiment one way or the other (e.g. "I am creating a Drupal theme today."). The greenish or reddish boxes show positive or negative sentiment (e.g. "I love/hate creating Drupal themes!"). I think that's how it works.
Hehe, this is fun. I searched for all our competition in the sentiment grid... the only one I found so far which is also swell is Wordpress. But they definitely should put Links behind the colored sqares. Like it is, you cannot find out how they arrive at their matrix.
This tools is fun but I doubt if it is really meaningful or representative. For instance when you look for the word "spam", it has an overall sentiment of "great". I would really like to see the calculations and results behind that one.
This post is interesting, as a lot of companies that are conducting social media marketing are using sentiment as a tactical tool and strategic device in customer engagement. Are you using this tool to integrate into your products at all?
Hi John. I think that something like this could be an interesting extension for Mollom, the technology that will power our content assurance engine at Acquia. Right now Mollom identifies spam. But if it could also judge sentiment, and give site/community owners a heads up when a thread sentiment hits extremes, it could help moderators engage more productively. Having a troll detector would be nice :-).
If you'd like to discuss it further, shoot me a note through the contact form above.