Let’s face it, keeping up with the latest on Windows Azure is hard. Whether it is a new feature announcement, a white paper, a code sample or just another attempt at “cloud washing,” it’s difficult to keep up with the latest, no matter how adept you are at mining the various social media channels.

This is why we built this “intelligent” twitter bot ( @AzureUpdates) as a weekend project. @AzureUpdates is designed to keep you to up to date with all things #WindowsAzure, in and around the Twitterverse.

Here’s how it works:

1) The bot looks for new tweets with the hashtag #WindowsAzure or #Azure *AND* have a (retweet_count + favorite_count) of five or more. Furthermore, we count retweets/favorites only if they are coming from a twitter user with a follower_count of 50 or more.

This way we know that we have a “tweet worth retweeting.” (Okay, fine, I admit I’ve been watching too many “ideas worth spreading”-themed TED talks lately.)

2) Since the #Azure hashtag is used by folks who are not just talking about Windows Azure (overzealous Azure developers may have likely forgotten that “Azure” also stands for a variation of the color blue), we apply a supervised machine learning algorithm based on Naïve Bayes classifier to filter out the tweets that are not referring to Windows Azure. (I am making it sound like I know a lot of machine learning when in fact I don’t. I only just completed a coursera course on machine learning. But if you are interested in this topic, I highly recommend it.)

3)  Now, some sneaky folks don’t retweet; they copy the tweet as a quote, modify it ever so slightly, and then tweet it. Now, we don’t want our twitterbot to be tweeting the same tweet again (remember our objective is to reduce the signal/noise ratio). Fortunately it’s not a tough problem to solve, at least for tweets that contain a hyperlink. Since we log every (re)tweet, we have an indexed inventory of all the “reduced”  hyperlinks. So if someone changes the text but points to the same “reduced” hyperlink, we will simply skip that tweet.

4) Okay, we know what you’re thinking. “I could use my super fancy twitter client, set up a search for the hashtag, language, set engagement counts for retweets, replies and favorite, and filter out unneeded tweets” Well, you certainly can, but who has time for that? Hopefully the sum of all the little “intelligent” features we described above is greater than the individual parts.

That’s it. Hope you follow @AzureUpdates.  Before we go, let me will give you one more reason to follow @AzureUpdates.  It is built on all-around favorite #WindowsAzure #MobileServices!

Thanks to Mohit Chabra from AIS for helping with the development of this bot.