|
|
1) in theory, you need training datasets to use classifiers or need to manually create rules. Probably rules are already effective as spams are usually posted by (1) new IDs, (2) similar/identical posts, (3) links to their target websites (this one is tricky tho).
2) actually, what can be done easily is to use the crowd wisdom. Assign a trustworthiness score to each user. When there are multiple reports of spam by several users, and their total trustworthiness exceeds a threshold, automatically move it to a quarantine area for Forum admin to process manually.
3) already banned certain ip and email addresses for user registration? can we not allow new users to post articles until some days after the registration?
just my 2c.
原帖由 ubuntuhk 于 31-3-2011 19:02 发表
谢谢羽天,我们会加大打击spam id的力度,不过只要论坛还是开放注册,就无法完全禁止这种行为,目前在技术上我们还没有自动分析spam的方案(其实和过滤spam mail挺类似,不过没有人来做这种论坛插件),我有实现这个 ... |
评分
-
查看全部评分
|