Apply babycart spam filtering to comments
Short, concise description of the idea
Offer users the ability to spam-filter their comments.
Full description of the idea
Periodically, users attempt to spam Livejournal users via comments placed in their journals. It would be useful if users, as well as community moderators, could opt to automatically pass some or all of the comments they receive through a spam filter. There would likely be account options of "spam filter all comments, just comments posted by non-friends, and just anonymous comments. (Another option might be "comments from users less than a week old.") Comments detected as spam would be rejected (or, optionally, automatically screened).
- Reduce the degree to which Livejournal is the victim of comment spam (which takes up storage, causes unnecessary messages to be sent via email, etc.).
- Reduce the hassle that users have to go through dealing with comment spam.
- Reduce the spam-related caseload for the abuse team.
- If this is a paid user feature, it is additional incentive to upgrade to a paid account.
An ordered list of problems/issues involved
- Some legitimate comments may be mistakenly classified as spam.
- Computational resources are consumed in checking comments when they are posted.
An organized list, or a few short paragraphs detailing suggestions for implementation
- Babycart -- http://www.cynistar.net/~apthorpe/code/babycart/ -- is a program which allows SpamAssassin's filtering to be incorporated into other software. This might make a good starting place for an implementation.
- False positives would be reduced by only look at the URIs within the comment, via SURBL (www.surbl.org). Since LJ spam generally depends on getting a user to go to another website, filtering on these URLs is effective. Also, this reduces the necessity of doing more heavyweight spam filtering checks.