Online Social Networks (OSNs) have become the most used Internet applications attracting hundreds of millions active users every day. The large amount of valuable information in OSNs (not even before available) has attracted the research community to design sophisticated techniques to collect, process, interpret and apply these data into a large range of disciplines including Sociology, Marketing, Computer Science, etc. This thesis presents a series of contributions into this incipient area. First, we present a comprehensive framework to perform large scale measurements in OSNs. To this end, the tools and strategies followed to capture representative datasets are described. Furthermore, we present the lessons learned during the crawling process in order to help the reader in a future measurement campaign. Second, using the previous datasets, this thesis address two fundamental aspects that are critical in order to have a clear understanding of the Social Media ecosystem. One the one hand, we characterize the birth and grow of OSNs. In particular, we perform a deep study for a second generation OSN such as Google+ (a OSN released by Google in 2011) and compare its growth with other first generation OSNs such as Twitter. On the other hand, we characterize the information propagation in OSNs in several manners. First, we use Twitter to perform a geographical analysis of the information propagation. Furthermore, we carefully analyze the propagation information in Google+. In particular, we analyze the information propagation trees and the information propagation forests that analyze the propagation information of a piece of content through multiple trees. To the best of our knowledge any previous study has addressed this issue. Finally, the last contribution of this thesis focuses on the analysis of the load received by an OSN system such as Twitter. The conducted research lead to the following main four findings: (i) Second Generation OSNs are expected to grow much faster that the correspondent First Generation OSNs, however they struggle to get users actively engage in the system. This is the case of G+ that is growing at a impressive rate of 350K new users registered per day. However a large fraction (83%) of its users have never been active, and those that present activity are typically significantly less engaged in the system than users in Facebook or Twitter. (ii) The information propagates faster but following shorter paths in Twitter than in G+. This is a consequence of the way in which information is shown in each system. Secuentialbased systems such as Twitter force short-term conversations among their users whereas Selective-based systems such as those used in G+ or Facebook chooses which content to show to each user based on his preferences, volume of interactions with other users, etc. This helps to prolong the lifespan of conversations in the OSN.(iii) Our analysis of the geographical propagation of information in Twitter reveals that users tend to send tweets from a sole geographical location. Furthermore, the level of locality associated to the social relationships varies across countries and thus for some countries like Brazil it is more likely that the information remains local than for other countries such as Australia. (iv) Our analysis of the load of Twitter system indicates that the arrival process of tweets follows a model similar to a Gaussian with a noticeable day-night pattern. In short the work presented in this thesis allows advancing our knowledge of the Social Media ecosystem in essential directions such as the formation and growth of OSNs or the propagation of information in these systems. The important reported findings will help to develop new services on top of OSNs.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados