Expertise Discovery in Decentralised Online Social Networks Sana Showkat Ara Insight Centre For Data Analytics, NUI Galway Galway, Ireland sana.ara@insight-centre.org Subhasis Thakur Insight Centre For Data Analytics, NUI Galway Galway, Ireland Subhasis.thakur@insight-centre.org John G. Breslin Insight Centre For Data Analytics, NUI Galway Galway, Ireland John.breslin@insight-centre.org ABSTRACT Distributed Social Networks (DSNs) are the solution to the privacy and security problems of online social networks. In DSN, a user controls their own data as it chooses personal storage for its so- cial network data. In absence of a centralized entity with access to all social network data, information retrieval becomes dicult in DSNs. In this paper we propose to use crowd sourcing for informa- tion retrieval in a DSN. We analyze a popular information retrieval problem called expert search in a social network. In this paper, we present an algorithm for such a crowd sourcing based search process which includes solution for (a) the worker selection prob- lem (b) the task selection problem and (c) the reward distribution problem. Using experimental evaluation, we show that, the search algorithms proposed in this paper can be as ecient as a greedy search algorithm with access to entire social network information. KEYWORDS Expertise Discovery, Decentralized, Online Social Network ACM Reference format: Sana Showkat Ara, Subhasis Thakur, and John G. Breslin. 2017. Exper- tise Discovery in Decentralised Online Social Networks. In Proceedings of ASONAM ’17, Sydney, Australia, July 31-August 03, 2017, 6 pages. https://doi.org/10.1145/3110025.3110048 1 INTRODUCTION Social networking has become the most popular communication medium of human life. However, centralised social networks have security and privacy problems. In a social network, users store large amounts of private data and they have very limited authority to control their own data. Social network provider have immense opportunity to extract, analyze user’s data anonymously and they may handover user data to third parties for personalised advertising. Therefore, the absence of condentiality and data ownership is the most discussed drawback of a centralized social network. Distributed Social Network (DSN) can provide a potential mit- igation of the security and privacy problems of OSNs. In DSN, a user chooses the storage location of its data and it controls the access to such storage. This helps to eliminate certain privacy and security issues. The benets of distributed social networks are open source code, decentralized, not owned by any central authority ACM acknowledges that this contribution was authored or co-authored by an employee, contractor or aliate of a national government. As such, the Government retains a nonexclusive, royalty-free right to publish or reproduce this article, or to allow others to do so, for Government purposes only. ASONAM ’17, July 31-August 03, 2017, Sydney, Australia © 2017 Association for Computing Machinery. ACM ISBN 978-1-4503-4993-2/17/07. . . $15.00 https://doi.org/10.1145/3110025.3110048 and as a result it gives back control of the data to users. In some decentralized social networks, user data remain in user machine [1, 2]. Information retrieval and search is dicult in DSN compared with OSN. In OSN, the social network provider has access to the entire social network data. In OSN, a query from a user can be eciently answered with sophisticated indexing techniques on the entire social network data. But there is no centralized entity in a DSN with access to all the social network data. Asknext [10] developed a protocol to search for an agent in OSN and its requires complete data of the OSN. The algorithms developed in this paper are ecient for incomplete OSN, specially decentralized OSN. In this paper, we assume that a search process in DSN will use the following procedure: • A user depends on its contacts in the DSN for an answer to its query, i.e., a user sends a query to its neighbours in a DSN who may have the answer and may know any other user who has the answer. • So, it sends the query to its neighbours and who may forward it to their respective neighbours. • For example, a user Sam is looking for a knowledgeable person in machine learning. Alice is another user of the social network who is unknown to Sam but they have a common friend Bob. Bob regards Alice as a machine learning expert though Alice’s activities in the social network. In this scenario, Bob will recommend Alice to Sam. Note that, in a OSN, this search could be answered by the OSN provider who has access to the entire social network information. But in a DSN, a user depends on its neighbors for such a search. Such a search initiated by a user may resemble a cascade on the network as the query spreads on the DSN. Such a search process can be very inecient. This is because, users may not participate in the search, users may not have the knowledge or expertise to answer the query or even route the query to the users who may have the answer. To overcome the above problems, in this paper we employ crowdsourcing. The proposed search process solves three problems. • Worker selection problem: We need to select workers who are most likely to answer the query or route the query to the users who has the expertise to answer it. • Task selection problem: We need to nd a suitable subset of the recommended neighbours of the workers to whom the the workers will propagate the query. • Reward distribution problem: We need to nd a payment scheme for the workers who participate in the search. The payment scheme should encourage truthful behaviour of the users. 2017 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 244