HUMBOLDT-UNIVERSITÄT ZU BERLIN
COMPUTER SCIENCE DEPARTMENT
Systems Architecture Group

Head:Prof. Dr. Jens-Peter Redlich
Secretary: Silvia Schoch
Phone:+49 (30) 2093 3400

 

     

ARM4SNS - Anonymous Reputation Management for Social Network Services

Contacts:  Prof. Jens-Peter Redlich, Dr. Wolf Müller, Henryk Plötz, Christian Carstensen. Martin Stigge.

Summary:

Social Network Services (SNS) help people to come together online around shared interests or causes. In former times the collaboration of people was restricted to their physical surrounding. Just recently evolving enhanced communication possibilities (characterized by low price, convenient user interface, feature-rich service infrastructure) allow people to form social networks even if they live in distant places of the world.

A good example for understanding the inherent potential of SNS is the migration of the traditional printing service into the online publishing services of the Web 2.0.

In times of classical print-media the content was generated by a limited number of dedicated authors. The content was evaluated (quality controlled) by editors or lectors.  The classical methods of information distribution were transformed into corresponding online versions (home pages, centrally operated web-portals, reviewed online publication, online encyclopedias, and search engines).  Within the Web 2.0 a qualitative change in the publication process occurs. Social Network Services offer a platform for ordinary people to use, and even contribute to the service. What distinguishes the Web 2.0 from the first generation web are the contributions of its spectacular user base – every user can become an author, no matter how specific his interests are.  Web 2.0 publishing services manage large quantities of articles that cover a huge variety of subjects at a fine granularity (special interests of a small fraction of people are presented by the activists of even that fraction).

The arising structures of Social Networks Services create masses of new content with great differences regarding their quality. Users are often overwhelmed by the amount of content and need help to decide which content (or generally which resources) to use/consume. Reputation systems are a technology with the potential to resolve this problem. A key issue within this context is privacy, which is critical to the acceptance of new Services, so reputation systems of the future should be able to incorporate the protection of privacy.

We propose an architectural design which fulfills these requirements with respect to the user's anonymity and privacy. We do this in an example case of a file sharing system, for which a Reputation Management System is developed. The proposed design provides a clear separation of information about the quality of the content and its origin.

For the example of a peer-to-peer file sharing service we have designed a system (Anonymous Reputation Management for File Sharing (ARM4FS).The ARM4FS system should enable users of file sharing systems to give feedback on their experiences with particular other users or files (more generally: resources). The system will keep track of this feedback information internally and might then be able to come up with recommendations for other resources based on the reputation information about their authors. Our system should derive reputation information from previous ratings about other documents published by the same authors.

The user in turn can take advantage of this information in order to decide which resource to use. A major goal is to keep all personally identifiable (or even pseudonymous) information sealed off inside the system, so that an outside attacker cannot correlate users and the resources they either provided or consumed. (At least the reputation management system should not make this easier than the resource-providing service (e.g. a file sharing system) itself does. Furthermore the user who provides feedback should stay anonymous. That means that even the author of the evaluated content must not be able to uncover the identity of this user.
 

It should be impossible to reveal the identity of the author of any given document. For two given documents it should even be impossible to find out if they originate from the same author or not.

Further Details:

Publications:

  • SAR-PR-2008-18
    Sybil Proof Anonymous Reputation Management.

    Wolf Müller1, Henryk Plötz1, Jens-Peter Redlich1, Takashi Shiraki2 (1Humboldt University Berlin, 2NEC Corporation), 10 pages, SecureComm 2008: 4th International Conference on Security and Privacy in Communication Networks, Istanbul, Turkey, 2008.
    [Abstract], [Conference Paper]


Legal disclaimer.   © 2018 Humboldt-Universität zu Berlin, Computer Science Department, Systems Architecture Group.Contact: sar@informatik.hu-berlin.de .