- Reputation system
A reputation system is a type of
collaborative filtering algorithm which attempts to determine ratings for a collection of entities, given a collection of opinions that those entities hold about each other. This is similar to arecommendation system , but with the purpose of entities recommending each other, rather than some external set of entities (such as books, movies, or music).Reputation systems are often useful in large online communities in which users may frequently have the opportunity to interact with users with whom they have no prior experience or in communities where
user generated content is posted likeYouTube orFlickr . In such a situation, it is often helpful to base the decision whether or not to interact with that user on the prior experiences of other users.Reputation systems may also be coupled with an incentive system to reward good behavior and punish bad behavior. For instance, users with high reputation may be granted special privileges, whereas users with low or unestablished reputation may have limited privileges.
Types of reputation systems
A simple reputation system, employed by
eBay , is to record a rating (either positive, negative, or neutral) after each pair of users conducts a transaction. A user's reputation comprises the count of positive and negative transactions in that user's history.More sophisticated algorithms scale an individual entity's contribution to other node's reputation by that entity's own reputation.
PageRank is such a system, used for ranking web pages based on the link structure of the web. In PageRank, each web page's contribution to another page is proportional to its own pagerank, and inversely proportional to its number of outlinks.Reputation systems are also emerging which provide a unified, and in many cases objective, appraisal of the impact to reputation of a particular news item, story, blog or online posting. The systems also utilise complex algorithms to firstly capture the data in question but then rank and score the item as to whether it improves or degrades the reputation of the individual, company or brand in question.
Practical applications
* Search: web (see
PageRank ), blogs (see blog search engines)
* eCommerce:eBay ,Epinions ,Bizrate
* Social news:Slashdot ,Reddit ,Digg
* Programming communities:Advogato , freelance marketplaces
* Internet Security:TrustedSource
* Email: anti-spam techniques, reputation lookup (RapLeaf )
* Peer-to-peer: identifying trusted nodesAttacks on reputation systems
A
Sybil attack is one in which an attacker subverts the reputation system by creating a large number of pseudonymous entities, and using them to gain a disproportionately large influence. A reputation system's vulnerability to a Sybil attack depends on how cheaply Sybils can be generated, the degree to which the reputation system accepts input from entities that do not have a chain of trust linking them to a trusted entity, and whether the reputation system treats all entities identically. It is named after the subject of the book "Sybil", a case study of a woman with multiple personality disorder.ee also
*
Reputation management
*Reputation service
*Collaborative filtering
*Web of trust
*Trust metric References
* [http://www.si.umich.edu/~presnick/papers/cacm00/reputations.pdf Reputation Systems] . P. Resnick, R. Zeckhauser, E. Friedman, K. Kuwabara. Communications of the ACM, 2000.
* [http://www.cs.rice.edu/Conferences/IPTPS02/101.pdf The Sybil Attack] J.R. Douceur. IPTPS02 2002.
* [http://www.ra.ethz.ch/CDstore/www2004/docs/1p403.pdf Propagation of Trust and Distrust] R. Guha, R. Kumar, P. Raghavan, A. Tomkins. WWW2004.
* [http://www.sigcomm.org/sigcomm2005/paper-CheFri.pdf Sybilproof reputation mechanisms] A. Cheng, E. Friedman. SIGCOMM workshop on Economics of peer-to-peer systems, 2005.
* [http://www.cs.ucl.ac.uk/staff/d.quercia/publications/quercia07lightweight.pdf Lightweight Distributed Trust Propagation] . D. Quercia, S. Hailes, L. Capra. ICDM 2007.External links
* [http://yury.name/reputation.html Reputation Systems] - Tutorial by Yury Lifshits
* [http://web.si.umich.edu/reputations/ Reputations Research Network] - a website from Michigan university
* [http://www.cs.cornell.edu/People/egs/credence/ Credence project] - Cornell project for p2p reputations
Wikimedia Foundation. 2010.
