toward spam 2.0: an evaluation of web 2.0 anti-spam methods
DESCRIPTION
Spammers have proven very powerfully adapt-able, if we thwart all current spam methods, they will find new loophole to use them. Blogs, comments, forums, opinions, online communities, wikis and tags are nowadays targets for their campaigns. This paper presents analysis of current anti-spam methods in Web 2.0 for spam detection and prevention against our proposed evaluation framework. The framework is a comprehensive framework to evaluate anti-spam methods from different perspectives. Our framework shows that the need for more robust methods which are prevention based, unsupervised and do not increase user and system interaction complexity is highly demanded. More info: http://debii.curtin.edu.au/~pedram/research/publications/83-toward-spam-20-an-evaluation-of-web-20-anti-spam-methods.htmlTRANSCRIPT
Toward Spam 2.0: An Evaluation of Web 2.0 Anti-Spam Methods
Pedram HayatiVidyasagar Potdar
Curtin University of Technology, Western AustraliaDigital Ecosystem and Business Intelligence Institute
Anti-Spam Research Laboratoryhttp://asrl.debii.curtin.edu.au
AGENDA• Introduction• Web 2.0 SPAM (SPAM 2.0)• Proposed framework• Evaluation• Conclusion
INTRODUCTION• Email spam is not the only campaign for
spammers as they always find new targets to achieve their desires.
• Web spam• Spammers nowadays, post promotional
comments on blogs, write advertisement reviews for products, reply online forums threads with junk content, create eye-catching user profiles on online community websites, manipulate Wiki pages, and create mislead ing tags for their documents
CONSEQUENCES• Tricking search engine to rank spam
and junk contents higher. Hence it decreases quality of search engine results.
• Misleading users to view unsolicited content.
CONTRIBUTION• We focus on current anti-spam
methods in– Blogs– Online forums– Wikis– Tags– Online communities
• Evaluate them against our proposed framework
Web 2.0 SPAM• Second generation of WWW• Easier platform to generate content
– Legitimate– Spam!
• Web 2.0 spam or Spam 2.0
Spam 2.0 is• Hosting blogs, writing blog
comments and making trackbacks
Spam 2.0 is• Posting new threads in online forums
Spam 2.0 is• Writing reviews/opinions
Spam 2.0 is• Creating user profiles in online communities
Spam 2.0 is• Modifying Wikis
Spam 2.0 is• Making tags
Proposed Framework• Is the method a detection strategy?• a. Is it a language dependent method?• b. Is the method content based or meta-data based?• c. Does the method use supervised, semi-supervised or• non-supervised machine learning approach?• d. Is the method behaviour based?• e. Does the method decrease user-interaction convenient?• 2. Is the method a prevention strategy?• a. Does the method prevent spammer to use user net-• work resources?• b. Does the method increase complexity of user-• interaction with system?
EVALUATION
• Blog, comment and trackback spam• Review/Opinion Spam• Social Spam• Wiki and forums spam
CONCLUSION
• Framework for evaluation of Anti-Spam 2.0 methods
• Supervised methods need up-to-date datasets
• Prevention method put pressure on user side and increase complexity of system
• Framework shows that there is need for– Prevention based methods– Unsupervised methods– Not increasing complexity of system
THANK YOU!
• http://debii.curtin.edu.au/~pedram/• http://asrl.curtin.edu.au/