Anders Andersen

Abstract

Njål Borch, Anders Andersen (Supervisor), Lars Kristian Vognild (Supervisor), Paul Havinga (Supervisor), The Socialized.Net - The design and implementation of a distributed social network, Institutt for informatikk (trykkeri: Tromsprodukt), Institutt for informatikk, 2007, ISBN 978-82-92461-76-1.

he information age has brought forward an ever growing amount of information and other resources, such as videos, pictures and music. Allowing users to efficiently locate interesting resources is of great importance. While several search systems exist, the best known being Google, these are typically not personalized. This means that resources are recommended by the same criteria for all users, not taking into account the user’s interests. This has led to an increased focus on personalizing services such as search and recommendation systems. At the same time, Peer-to-Peer networks have demonstrated a supe- rior technology to the traditional server based approach for content distribution and local communication. Peer-to-Peer networks are very robust, scales to large amounts of computers and are robust to failure of individual machines. These networks also allow end users to share large files at virtually no cost, paving the road for global distribution of user generated content. Current Peer-to-Peer networks are however typically either created for global or local (ad-hoc) net- work use. They are often lacking in personalization, in controlling diverse interests and in their handling of malicious nodes and network pollution. Together, this can severely limit the usefulness of Peer-to-Peer networks for large scale content sharing. The Socialized.Net presented in this work is an implementation of a fully distributed social network. The Socialized.Net investigates utilizing social bonds to provide a highly scalable, dynamic, robust and personal search platform. The Socialized.Net also bridges local networks, such as wireless ad-hoc networks, and the Internet. This allows for locality aware applications and local co-operation without a need for additional protocols. The Socialized.Net search platform has been integrated with a set of highly diverse applications with various needs. This has demonstrated the feasibility of a distributed social network. The social network has also provided powerful additions to the test applications.