www.twitblock.org

I’ve finally got round to building the Twitter app I’ve been thinking about for months. While everyone else is preoccupied with making fun, or cool apps, I’ve been thinking about the increasing problem of spam and junk followers on Twitter. I won’t go into why I think this is such a problem right now, plenty of time for that later.

This is just a quick announcement to say that I’ve released an early alpha version of a tool that I hope to develop into something genuinely useful. Currently it’s a simple scanner that analyses your followers for signs of “spammy” behaviour. I’ll post more details about these indicators soon, and I’ll also share some of the interesting discoveries I’ve been making about Twitter spam as I go on my mission.

UPDATE: I have posted about these indicators

Data mining for good, not evil

One of the principal aims of TwitBlock is to gather data in order to improve the service – i.e. to make it accurate enough that it could [in theory] be used to automatically filter spam out like an email junk filter endeavours.

By logging into TwitBlock (via Twitter OAuth of course) you are sharing the list of people that you block. As long as the app is authorized I can update this list and the app can learn from it.

Additionally I will be writing various bots (crawlers) that analyse Twitter activity in terms of suspicious behaviour and mine more data. More about these bots later too