Comments on: Cutting Back on DM Spam

By: Si Dawson

Si Dawson — Sun, 13 May 2012 04:18:00 +0000

Oh awesome! Thank you.

By: Joshua Mango

Joshua Mango — Sun, 13 May 2012 04:06:00 +0000

Thanks that’s a great tip! I’ll write a blog about it and link to this article.

By: Si Dawson

Si Dawson — Sun, 13 May 2012 03:58:00 +0000

The reason that’s happening is because the person sending them to you has had their account hacked. They need to change their password & check their Twitter | Settings | Applications (http://twitter.com/settings/applications) – revoke access to any that seem dodgy.

Unfortunately, the only way to stop them coming through is to unfollow the person in question.

By: Joshua Mango

Joshua Mango — Sun, 13 May 2012 03:50:00 +0000

Thanks! Any way to stop the ‘someone is posting nasty pictures of your mom’ (or something like that) DMs?

By: Eloísa Valdes

Eloísa Valdes — Mon, 30 Apr 2012 11:59:00 +0000

Oh. I see. It’s a really tricky one indeed. It’s nice to have an idea of the amount of work involved in such filtering and how you work with your algorithm. :) How do you gather that kind of data? I want to help you on that, but not sure how could I do it. The only thing that comes to mind is making an online survey, but even that doesn’t strike me as much effective. :x

By: Si Dawson

Si Dawson — Mon, 30 Apr 2012 00:31:00 +0000

Scanning bios is going to be a little unreliable – too easy to change a word here or there & side step any filters. There’s also not enough information to draw reliable conclusions.

Eg “My name is cam!” (with bad capitalisation) or “I love Ontario’s rivers and I love using my cam. Add me at pinterest” – both harmless, but tricky to distinguish from spam.

It’s generally easier to look at overall behaviour – you have more data to work with, so it’s less likely to give false positives. The last thing you want to do is flag an enthusiastic nature photographer as a porn bot.

Another issue is – for example, I follow someone whose icon is a pic of them in a bikini. They’re a personal trainer – so it’s totally appropriate, but could easily be mistaken for a porn bot if you ignored their content.

I’l continue to think about it. There may be a combination of bio+behaviour that might be useful – although whether that’s more indicative than merely behaviour I’ll need more data to assess.

Lots of factors to consider, as you can see.

By: Eloísa Valdes

Eloísa Valdes — Mon, 30 Apr 2012 00:04:00 +0000

Ok, I have some examples, from memory, of that kind of tweeters. You can identify them, not only through their avatars, but also through their “catch-phrases”, on their bios and tweets.

Examples:

“Any guys wanna tell me what to do on cam? add me at…”
“Let’s have some fun!!!…”
“Yet another Christmas alone… anyone single? add me at..”

Not sure this helps. :

By: Si Dawson

Si Dawson — Thu, 26 Apr 2012 01:28:00 +0000

Wow. So many questions. ok:

1. Where it says “allow acces” and “revoke access” then yes, do that. The other cases you won’t need to.

2. If you revoke access to something (eg Seesmic) you can always allow access again later. It’s just a switch, on, off, on, off. No problem at all.

3. “revoke access” is the phrase used by Twitter on your applications page (under Twitter | Settings | Applications – the menu is at the top right)

4. If (above) I’ve got a link “here”, then click that. Where it takes you will depend on what’s needed to deal with that specific form of DM spam. In other words, each url is different. If you’re not comfortable, then either don’t click, or hover over the link to see the url before you click.

5. Do as many as you like – but the list is sorted in order of badness. So, even just first one will help a lot. The others help too, but less as you go down the list.

I think that’s everything. If you’re finding it confusing, you can always just delete the DMs as they come in. It’s tedious, but that works too.

By: Skakk4

Skakk4 — Wed, 25 Apr 2012 14:42:00 +0000

Do you mean to follow these applikations request and sign up and join and follow them just to unfollow so they cant harm you. Another thing I am new to twitter and I’m not finished and ready to use seesmic cause I should verify my twitter account and there is problems but under twitter connection settings I did revoke acces to seesmic. Is that a mistake. I have difficulties about understanding the word revoke so it crashes all communication. Do you mean to press the link “here” so I will be taken to twitter to verify my account. Do you want me to do so in all cases above under the title Cutting back on DM spam. I’m working hard to understand so many different sensory impressions and also problems with the language. Twit Cleaner has done an awesom job helping me and I do appriciate the help <3

By: Si Dawson

Si Dawson — Sat, 21 Apr 2012 21:25:00 +0000

If they’re “porn tweeters” due to their image, I’d only be able to identify them with relatively sophisticated image analysis (a little beyond the scope of Twit Cleaner).

If it’s their tweets (NOT DMs, since I don’t look at those), I may be able to do something about it.

Additionally, right now Twit Cleaner is aimed at those you’re following (ie, you unfollow to remove the problem, not block) – other side of the equation.

That said: if you can give me three or four examples, I may be able to find a common thread & create a new category.