Twitter Anti-Spam Project Introduction & TweetDeck Anti-Spam Functions

Twitter is one tweetdeck_anti_spamof the premier social media applications, but due to it's size and open nature it is also one of the easiest to spam. The very open community nature of Twitter and it's feed system have been exploited by spammers time and time again.

Twitter has some tools, filters and options in place to help you fight it, but just like anything in life nothing is foolproof. In today's video I'm talking about my idea for a Twitter Anti-Spam Project (outlined below the video) and how you can utilize TweetDeck's newest filter features and a few of the existing ones to keep yourself safe from Twitter SPAM.

But take this with a grain of salt: This is not a fool-proof system and it can result in false-negatives and false-positives. But it's definitely better than nothing at all.

In the video you'll find out how to use TweetDecks features to filter out SPAM and CRAP.

Twitter Anti-Spam Project Proposal:

My idea is to have a central filter system which is not directly on the twitter system. If it were it would but significant strain on Twitters systems and plus, there are plenty of legitimate Tweets coming through that might appear spammy at first. Twitter cannot and will not take such a position to go around blocking random tweets.

Thus we need a separate system that will allow complete user level control with well defined filters. Here are my basic requirements for the system:

  • Be separate from the Twitter core
  • Have an identical API to Twitter so that any existing applications would only have to use our servers address instead of Twitters to pull FILTERED tweets from
  • Ability to filter by:
    • Names
    • Locations
    • User rating
    • Tweet source: api/web/tweetdeck (more on this one below)
    • Profile image type
    • User profile scores
      • Tweet rate
      • Tweet types: RT? all quotes? gibberish? all sales with link tweets? ect..
      • Link in tweet or not
      • Link spam detection: check link for automatic anti-spam linking techniques (ie, adding useless trailing characters to bit.ly links to make them appear unique)
      • Following to Followers ratio
      • Profile age
      • Profile spam rating
      • Profile usage rating (kind of like what those twitter sites do where they give profiles a rating from A+ to F-)
    • Link spam detection (same as the one above under profile but more general setting for all links)
    • other things not thought of yet…
  • User contribute to overall ratings of other users and spam is filtered out through the help of everyone. Similar to what Gmails spam detection does. (Hm, hey Google, can we get a copy of your spam protocol to help us here? 🙂 )
  • Ability to report faulty detections
  • Per user settings which can override global settings
  • Follower vetting
    • Automatic
    • Manual
    • Semi-manual
  • other things as well that haven't been thought of yet.
  • I have written this Bookmarklet to remove the users in search.twitter.com, who you don’t want to have in your results.

    Write the names of users, you want to get filtered out, in lower case seperated by comma in the field, where you find now username1 and username2.

    Maybe someone with better skills than me can write a browser extension, which would be more comfortable. 🙂

    This is the code for the bookmarklet:

    javascript:
    {
    var blist = ‘username1, username2?;
    var banned = blist.split(‘, ‘);
    var i;
    var j = document.getElementsByTagName(‘a’).length;

    function removeBanned(nr)
    {
    var element=document.getElementsByTagName(‘a’)[nr];
    var x;
    var klein;

    klein=element.text.toLowerCase();
    for (x= 0;x<banned.length;x++)
    {
    if(klein==banned[x])
    {
    element.parentNode.parentNode.parentNode.removeChild(element.parentNode.parentNode);
    i–;
    }
    }
    }

    for (i = 0;i < j;i++)
    {
    if(document.getElementsByTagName('a')[i].className=='username')
    {
    removeBanned(i);
    }
    }
    }

  • >