Creating social networks to improve peer-to-peer networking

Abstract

We use knowledge discovery techniques to guide the creation of efficient overlay networks for peer-to-peer file sharing. An overlay network specifies the logical connections among peers in a network and is distinct from the physical connections of the network. It determines the order in which peers will be queried when a user is searching for a specific file. To better understand the role of the network overlay structure in the performance of peer-to-peer file sharing protocols, we compare several methods for creating overlay networks. We analyze the networks using data from a campus network for peer-to-peer file sharing that recorded anonymized data on 6,528 users sharing 291,925 music files over an 81-day period. We propose a novel protocol for overlay creation based on a model of user preference identified by latent-variable clustering with hierarchical Dirichlet processes (HDPs). Our simulations and empirical studies show that the clusters of songs created by HDPs effectively model user behavior and can be used to create desirable network overlays that outperform alternative approaches.

Examining spectral space of complex networks with positive and negative links

“…To understand and utilise the information in a social network, researches have developed various methods to capture the structure and characteristics of the network from different perspectives (Kleinberg, 2007;Shiga et al, 2007;Tantipathananandh et al, 2007;Backstrom et al, 2006;Koren et al, 2006;Berger-Wolf and Saia, 2006;Kumar et al, 2006;Fast et al, 2005;Spertus et al, 2005;Baumes et al, 2004;Girvan and Newman, 2002;Kempe et al, 2003). Among them, spectral analysis of the adjacency matrix and its variants (e.g., Laplacian matrix and normal matrix) has shown intimate relationship between the combinatorial characteristics of a graph and the algebraic spectral properties of its matrix (Seary and Richards, 2003).…”

Strategic reasoning about bundling in swarming systems

“…There has been a vast amount of work on P2P systems focusing on their performance analysis, e.g., [27], [18], content search, e.g., [9], [23], neighbor selection, e.g., [22], [5], replication strategies, e.g., [7] and legislation, e.g., [26]. However none of these deal with the problems of when and how to bundle content.…”

A recall-based cluster formation game in peer-to-peer systems

“…Similarly, [5] assumes predefined classification hierarchies based on which queries and data are categorized. Instead of predefined categories, [8] uses a learning approach that based on generalizing the shared data, learns the semantic categories they belong to and then uses those for clustering. Clustering in [15] is based on the schemes of the peers and on predefined policies provided by human experts.…”

“…While, there is a large body of research on the discovery and construction of clustered overlays [3,5,15,21,11,8,4,6], their maintenance, which is imperative for coping with the dynamic nature of peers, has been mostly ignored.…”