Re: [twitter-dev] Re: Provide a spam score (or rathe r a “good citizen” flag)

2010-10-17 Thread Adam Green
Terms like spammer or good citizen are kind of vague. What are
your goals in ranking users? Are you looking for good people to follow
or engage with? In that case it doesn't really matter if they are a
bot or a human. What matters is whether others find them interesting.
Even more important is how interesting they are within a specific
domain. The best sports account, whether their tweets come from a
script or a human typing, means nothing to someone who doesn't follow
sports. Even if Twitter HQ maintained an absolute rank, it wouldn't
apply to all readers of an account. What I do for clients is rank
users within their specific area of interest based on a collection of
keywords. I ignore follower count, since that is so easy to game with
brute force methods. Mentions work as a better measurement of
popularity.

I have a detailed tutorial on mention tracking on my site:
http://140dev.com/twitter-api-programming-tutorials/identifying-influential-twitter-users/
Here is a summary of the algorithm. What I do is aggregate tweets for
keywords, and record @mentions. If @fred mentions @sally, I record
@fred as the source and @sally as the target. Then I can rank users by
the number of mentions they receive as the target, or even more useful
is ranking users by the number of different sources mention them. That
is hardest to game. The real key is that this is based on mentions in
tweets that contain specific keywords.

On Sun, Oct 17, 2010 at 6:12 AM, Dave-twiends i...@davesumter.com wrote:
 I think one solution is if twitter could use their hidden karma score
 to provide good citizen flag for us.

 So, they shouldn't provide any kind of score to the outside world via
 the web or API, but they could use it to provide an indicator for a
 clean account.

 In other words, we would assume all accounts to be potential spammers
 until we see the good citizen flag on the account, then we can be a
 little more trusting of that account. Not completely trusting, but
 just use it as an indicator..

 On Oct 17, 6:43 am, M. Edward (Ed) Borasky zn...@borasky-
 research.net wrote:
 I don't know about a karma score, but Twitalyzer does have an API
 and so does Klout. For that matter, Viralheat has an API and they can
 get both Twitalyzer and Klout scores.

 That said, I don't know that there's ever really going to be a one
 size fits all Twitter user metric. But there are quite a few
 crowdsourcing and curation tools starting to show up, some of them
 open source. But personally, I think it's more fun to just collect raw
 data via the API and roll your own. ;-)
 --
 M. Edward (Ed) Boraskyhttp://borasky-research.nethttp://twitter.com/znmeb

 A mathematician is a device for turning coffee into theorems. - Paul Erdos

 Quoting Justin justin.carl...@gmail.com:



  Rating/scoring users is something I'm working on as well and I agree.
  I've found sorting out bots and pure spammers to be very difficult.
  Some folks tweet so much they resemble bots/spam.

  Feels like a pipe dream but if they can I'd love a karma scoring
  system directly from the API.

  On Oct 16, 4:28 am, Dave-twiends i...@davesumter.com wrote:
  I really like the option of reporting spam via the api. I’ve been
  blocking spam on my site for a long time but this gives me an option
  to report it now, and hopefully get these account suspended quicker so
  that they don’t come back.

  It would be really great if we could have a proactive api function as
  well, where we could get the likelihood of a user being a spammer. I
  know this is really difficult to do, and wrought with pitfalls, but
  perhaps it could be structured in some way? Maybe you could provide a
  “good citizen” flag for a user (i.e a user that hasn’t had any
  complaints, and has a certain account age). That way you don’t
  negatively impact any users, but we can then at least treat these
  users differently when they sign up with our site.

  I’m seeing more and more the need to provide different limits to users
  based on private trust/karma score we develop for each user. This
  would be a very valuable input for us to detect potential problem
  users before they can cause damage.

  Thanks
  Dave

  --
  Twitter developer documentation and resources:http://dev.twitter.com/doc
  API updates via Twitter:http://twitter.com/twitterapi
  Issues/Enhancements 
  Tracker:http://code.google.com/p/twitter-api/issues/list
  Change your membership to this group:
 http://groups.google.com/group/twitter-development-talk

 --
 Twitter developer documentation and resources: http://dev.twitter.com/doc
 API updates via Twitter: http://twitter.com/twitterapi
 Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list
 Change your membership to this group: 
 http://groups.google.com/group/twitter-development-talk


-- 
Twitter developer documentation and resources: http://dev.twitter.com/doc
API updates via Twitter: http://twitter.com/twitterapi
Issues/Enhancements Tracker: 

Re: [twitter-dev] Re: Provide a spam score (or rathe r a “good citizen” flag)

2010-10-16 Thread M. Edward (Ed) Borasky
I don't know about a karma score, but Twitalyzer does have an API  
and so does Klout. For that matter, Viralheat has an API and they can  
get both Twitalyzer and Klout scores.


That said, I don't know that there's ever really going to be a one  
size fits all Twitter user metric. But there are quite a few  
crowdsourcing and curation tools starting to show up, some of them  
open source. But personally, I think it's more fun to just collect raw  
data via the API and roll your own. ;-)

--
M. Edward (Ed) Borasky
http://borasky-research.net http://twitter.com/znmeb

A mathematician is a device for turning coffee into theorems. - Paul Erdos


Quoting Justin justin.carl...@gmail.com:


Rating/scoring users is something I'm working on as well and I agree.
I've found sorting out bots and pure spammers to be very difficult.
Some folks tweet so much they resemble bots/spam.

Feels like a pipe dream but if they can I'd love a karma scoring
system directly from the API.




On Oct 16, 4:28 am, Dave-twiends i...@davesumter.com wrote:

I really like the option of reporting spam via the api. I’ve been
blocking spam on my site for a long time but this gives me an option
to report it now, and hopefully get these account suspended quicker so
that they don’t come back.

It would be really great if we could have a proactive api function as
well, where we could get the likelihood of a user being a spammer. I
know this is really difficult to do, and wrought with pitfalls, but
perhaps it could be structured in some way? Maybe you could provide a
“good citizen” flag for a user (i.e a user that hasn’t had any
complaints, and has a certain account age). That way you don’t
negatively impact any users, but we can then at least treat these
users differently when they sign up with our site.

I’m seeing more and more the need to provide different limits to users
based on private trust/karma score we develop for each user. This
would be a very valuable input for us to detect potential problem
users before they can cause damage.

Thanks
Dave


--
Twitter developer documentation and resources: http://dev.twitter.com/doc
API updates via Twitter: http://twitter.com/twitterapi
Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list
Change your membership to this group:   
http://groups.google.com/group/twitter-development-talk






--
Twitter developer documentation and resources: http://dev.twitter.com/doc
API updates via Twitter: http://twitter.com/twitterapi
Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list
Change your membership to this group: 
http://groups.google.com/group/twitter-development-talk