Scalable Link Prediction in Social Networks Based on Local Graph Characteristics

Abstract

Abstract-Online social networks (OSNs) like Facebook, Myspace, and Hi5 have become popular, because they allow users to easily share content or expand their social circle. OSNs recommend new friends to registered users based on local graph features (i.e. based on the number of common friends that two users share). However, OSNs do not exploit all different length paths of the network. Instead, they consider only pathways of maximum length 2 between a user and his candidate friends. On the other hand, there are global approaches, which detect the overall path structure in a network, being computationally prohibitive for huge-size social networks. In this paper, we provide friend recommendations, also known as the link prediction problem, by traversing all paths of a bounded length, based on the "algorithmic small world hypothesis". As a result, we are able to provide more accurate and faster friend recommendations. We perform an extensive experimental comparison of the proposed method against existing link prediction algorithms, using two real data sets (Hi5 and Epinions). Our experimental results show that our FriendLink algorithm outperforms other approaches in terms of effectiveness and efficiency in all data sets. Finally, we discuss extensively various experimental considerations, such as a possible MapReduce implementation of FriendLink algorithm to achieve scalability.

Domain Knowledge-Based Link Prediction in Customer-Product Bipartite Graph for Product Recommendation

“…30,31 and (3) structural similarity method. 1,32 This kind of classi¯cation method for modeling re°ects the di®erent ideas of the link prediction problem. The complexity of the maximum likelihood estimation method is higher, which is not suitable for the application in large network.…”

DPM: A novel distributed large-scale social graph processing framework for link prediction algorithms

“…the UserTask may generate results for some vertices, or even no results at all). Once the execution is completed, the current sub-results are partitioned (3) and sent to their corresponding Workers (4). Naturally, those sub-results belonging to the current Worker are stored locally.…”

“…Methods for link prediction use topology-based similarity metrics that can be categorized into path-based, neighbor-based (neighbor-based can be seen as a special case of path-based algorithms of length two) and random walk-based. Several social network recommendation algorithms based on these notions can be found in the literature [1, 3,4].…”

The layered structure of company share networks

“…The dividend problem is introduced in this paper. Rubin's algorithm [14], which is used at the core of the solution, has been exploited for finding paths connecting an individual in an online social network with friends of a friend at distance l [11]. However, density of social networks makes the approach unfeasible even for small values of l. On the contrary, sparsity and layered structure of ownership networks, allow for an efficient solution of the dividend problem.…”