Can Mollom detect libelous comments
Most of what Mollom does seems ro be related to spam. But, we are also worried about users commenting that "this guy is a liar" or "she's a whore", etc and it bogs down our whole comment process. Can Mollom help out with that kind of thing?
The current version of Mollom
The current version of Mollom allows you to report comments as profanity, but it won't block libelous comments yet. We're still working on that. The more people install Mollom and that report libelous content, the sooner we can try to block it.
Libelous comments
Folks, coming from a legal background and as a lecturer in Legal Regulation for Information Systems I'd humbly suggest that there is no automated tool that will flag or block libelous comments.
There are many reasons for this, the most significant being that there are a variety of legal systems to be addressed (the law in the US is different to the law in the Irish Republic, is different to the law in the UK, is different to the law in Australia, is different to the law in Belgium). At its simplest, a libel is a statement that is published which lowers (or is likely to lower) a person in the eyes of society. This requires additional context (for example background information on a person mentioned, their professional situation etc.) which might not be available to even the smartest machine learning tools. Unfortunately life is not like Star Trek.
Having effective measures to let people report libelous comments and then removing those comments quickly (and apologising to the person so affected) is the best approach (but will not absolve you of liability). If a particular commenter to your site is repeatedly posting libelous comments then block them from the site using blacklist functionality that is in most CMS.
This is one area where relying on a pure technology solution will inevitably land you in trouble as the legal responsibility is on you as a 'publisher' of the site to make sure that any content that might libel someone - telling the judge that "the software didn't catch it" will NOT be a defence.
While this is not legal advice (and shouldn't be construed as such in any circumstances) I'd be very wary about relying on an automated process to block libelous comments and I'd be very wary as a software developer about holding out my solution as being able to block libelous comments.
this link: http://www.tuppenceworth.ie/blog/index.php/2008/04/01/internet-libel-again/ has links from it to some resources that explore the challenges of libel on the internet.
Since this system is meant to
Since this system is meant to prevent spam or other malicious content from being posted I see no problem here.
Where is the difference between a libel being posted, then reported and deleted manually
- or -
eventually tuning the system so that it can catch them or prevent them. And if it doesn't catch them all then you still have to moderate much less then before.
I can imagine a few approaches here. I'll give one example:
1. catching words that are potentially libelous.
2. marking them as "unsure"
3. based on knowing they contain potentially harmful language, presenting the author not only with a captcha but with a legal warning e.g.: "Your comment contains potentially abusive language. Please make sure to follow our guidelines for posting comments to avoid (legal?) consequences."
4. if the comment still is getting posted offer, as you generally should for any user created content, a moderation/report link.
I see no potential harm in such a system, only benefits. However, you are correct that human moderation will most likely still be needed (since posters with those intentions will most likely ignore the warning) but from the legal side you have even better cards then before I'm sure. After all they ignored the warning again and CONFIRMED that they are willingly ignoring it along with your policy. Which should put them in a very bad situation to begin with and any lawyer should (will?) be happy to follow the provider logs.
I'm planing on using Mollom on a non-profit-organisation website (based on Drupal) and if things go smoothly I'll definitely promote this system. It might be a bit in the future but I'd love libel filtering since I assume my project is potentially targeted by huge amounts of libel comments.
Greetings from Germany,
user.