Bilkent University
Department of Computer Engineering


Cluster-based Profile Matching across Online Social Networks


Anisa Halimi
MS Student
Computer Engineering Department
Bilkent University

In today's world online social networks have become a popular way of communication. Users share about their social and professional life, hobbies, diseases, friends and opinions. The tremendous amount of data shared in these social networks can reveal relevant information about their users. Therefore, it is of significant interest to measure the privacy leak. We propose a model that will match the profiles of the users in different social networks by only using the publicly available data in these networks. Our model consists of three main steps: 1) computing the similarity between each pair of users within the social network and then separating the users in clusters by applying a naive approach or k-means; 2) matching the clusters obtained in one social network to the clusters obtained in the other social network; and 3) matching the user profiles within the matched clusters. To evaluate our model we use Foursquare and Twitter datasets.


DATE: 07 November, 2016, Monday @ 16:00