[ckan-discuss] Flagging SPAM

Rufus Pollock rufus.pollock at okfn.org
Thu Sep 8 00:44:29 BST 2011


On 2 September 2011 13:46, Pablo Mendes <pablomendes at gmail.com> wrote:
>
> David,
> Thanks for the prompt response.
>>
>>  To be honest, although we need to clean up spam users, we
>> are much more interested in cleaning up spam packages.

Just to say that we have now implemented recaptcha for user signups --
live on thedatahub.org already :-)

> Exactly, me too. But if a spammer stays alive, it can create more packages.
> If the spam package creation is automatic, and the spam package flagging is
> manual, then we have a problem.

At the moment we seem to have relatively few issues with spam creation
or editing of datasets -- it does happen but in general thanks to the
efforts you Pablo and others we are on top the problem.

> By flagging users I intend to block them from creating new packages. This
> would force spammers to create a new user and give us some time to breathe.

At the moment most spammers only do one edit with a given user account it seems.

> By the way, do we have a captcha for user creation?

There is one now!

> I also think that an incremental approach could work very well here. Start
> with a "Flag Spam" button that at least blocks packages from the front page.

Agreed. Note that the front page is about to dramatically change as we
have been working heavily on a new theme.

> Subsequently we can use the information gathered to train automatic spam
> detectors. This is related to some work I do, and I would be glad to
> implement the classifiers if you give me some training data (for example
> collected from the spam button).

That's a nice idea. One could hook nicely into the package save hook
in an extension.

Rufus



More information about the ckan-discuss mailing list