Tackling a novel problem of scalable secure distributed computing, this paper introduces a framework to address the processing of information from social networks in a new way. The authors “argue for a decentralized approach” that differs from the current common social networking architectures, relying on a single authority to keep the data. They envision a scenario “where the participants in the social network keep their own data [locally] and perform computations in a distributed manner without any central authority,” and exemplify for the first time in the literature “what can be computed in a large-scale social network while accounting for the human and social [sides] of its users.”
The paper defines the problem of computing in social networks as an S3 problem. The S3 problem involves scalability (for example, the complexity of the distributed computation), security (for example, accuracy and privacy), and the social aspect. The paper proves that the S3 problem can be reduced to “computing component-wise addition of vectors of integers with exactly one nonzero component, which is equal to one.”
Consequently, the paper presents a distributed protocol that solves the problem for several aggregation functions, based on “(1) a structured overlay where nodes are clustered into groups, (2) a homomorphic secret sharing scheme that allows the nodes to obfuscate their inputs, and (3) a verification [process that] tags the profiles of suspected nodes.”
Presenting a very interesting and current problem with a novel solution, and explaining it in a very detailed and technically sound manner, this paper is good reading for scholars and especially for engineers interested in processing information from social networks and in the future of distributed computing.
