• Undocumented features, Quarantine

      Thursday, 20 Sep 2007 - 13:59 GMT

      A few months ago we were working a new system for combating Spam content going into Connotea. What we used to do is delete accounts that violated our feeling for what we thought people should be doing on connotea. We would send an automated email to the deleted account that said:



      Thanks for getting back to me, and I’m sorry if our original email wasn’t clear.

      Your account was deleted by us during a crackdown on people using Connotea for marketing and other commercial purposes. Connotea is a service designed for scientists and clinicians to manage their academic references, and while we do not expect every link to be academic, we do take exception to people using Connotea solely to market products or to drive traffic to their own websites. I’m afraid your usage fell into this category, which is why we have deactivated you. I’m sure you will understand that the pollution of Connotea with these commercial links has a significant impact on the other users of the service.

      If you would like to use Connotea for academic purposes, then you are, of course, very welcome to open a new account. However, if you wish to use it to drive traffic to other websites, then I’m afraid we can’t help you, and I suggest you look for another social bookmarking website where this kind of use is acceptable. In a nutshell, it is fine if you are using Connotea to organise your own links so that you can find then again. What is not fine is using Connotea so that others can find links.

      I hope this is clearer to you now: please let me know if you have any further questions or comments.

      Thanks,

      Well, now we have taken a different approach. Instead of removing accounts we are just making them invisible, or quarantining them. The person owning the account can still log in, see their bookmarks, and post to the account, but no one else can see them. Their contributions don’t get included in searches or listed in lists by tags. There are a number of reasons for this. It means that if we make a mistake in identifying an account that that is bad, then fixing it is really really easy, we just life the quarantine. Instead of alerting spammers to the fact that they have been nobbled some just continue to post into a location that does not help with increasing google rank or attracting attention to specific sites or products.

      We have not documented this feature before but I wanted to mention it here in the context of another topic, terms and conditions. We don’t currently have any for Connotea, but it came up recently when I was contacted by an organization that is interested in investigating using Connotea to help index it’s scientific collection. They need to have a terms and conditions in place, so that they have some degree of confidence in what they are doing with the system. We have always taken an open approach to people’s data, with the API you can extract everything you put into Connotea, but now it is time to formulate a TOC that codifies the feeling that we have been working on up to now. I’ve started looking at some being used by similar sites, and I’ll post a preview here of what I am thinking about soon.

      As ever any comments from the community are welcome.

      - Ian

      Last updated: Thursday, 20 Sep 2007 - 13:59 GMT

      • Comments

        • Date:
          Friday, 02 Nov 2007 - 00:23 GMT
          Robert Muetzelfeldt said:

          Well, I understand your approach, but it’s a bit of a pain. I’ve just added a perfectly innocent bookmark, and had it quarantined, whereas 3 others were added with no problem. I tried deleting and re-adding it (and having typed in the graphical code), but still quarantined. Some re-jigging of this system is required.

        • Date:
          Sunday, 11 Nov 2007 - 11:49 GMT
          Matthew Loxton said:

          Would love to know what triggers a quarantine. I just added a cluster of citations imported from EBSCO, and those triggered a Captcha challenge, and were also quarantined. Does this mean that anytime I import a list of citations rather than one at a time, they will get quarantined?

        • Date:
          Tuesday, 13 Nov 2007 - 11:24 GMT
          Ian Mulvany said:

          Hi Matt, Robert,

          We have a number of rules that we use to test for Spam. Things like checking specific words, IP addresses, but also posting behavior. One of the things we noticed about human spam was that for a while spammers were adding a lot of posts within a very short time of one another, so we put in a test to look for a lot of posts being added in a short space of time, it seemed to make sense. It has since become clear that the system does not distinguish this from a file upload. This was a mistake on our part and we will fix this.

          - Ian


Search blogs

web feed Want a blog?

Submit this post to

Advertisement