Dissertation/Thesis Abstract

Encouraging User Engagement with Online Social Networks
by Lang, Juan, Ph.D., University of California, Davis, 2011, 111; 3499459
Abstract (Summary)

Online social network operators are interested in promoting use of their sites, because usage translates to revenue. In this work, we investigate how various factors in online social network user behavior correlates with the users' lifetimes in the online social networks. Many of the primary results are perhaps expected: the most popular users are more likely to remain online; users who are contacted early in their lifetimes are more likely to remain active in the online social network; and users are more likely to remain online if their online friends post frequently. Some results merit further investigation: for example, users who reveal more about themselves are more likely to remain online longer. This finding has implications for further study, e.g. into the impact of privacy settings on users' sociability. Some results are surprising: for example, users who follow only the most popular users are less likely to remain online long after creating their profiles than are users who follow at least one less popular user.

Because online social network operators may wish to encourage users to post more frequently, they may offer incentves for doing so. This work also shows that, given incentives, users may be more inclined to manipulate other users in order to increase their chances of receiving the offered incentives. In many cases, such manipulation is possible to detect automatically, perhaps allowing social network operators to deter users from engaging in such manipulation.

Finally, because the most popular users have the longest expected lifetimes in online social networks, a natural question to ask is how the most popular users become so popular. One very common model for social network growth is preferential attachment, in which existing nodes gain edges from new nodes in proportion to their existing degree. In this work, we show evidence for a surprising form of edge formation in which existing nodes add edges to newly joining nodes, which the new nodes may reciprocate.

Indexing (document details)
Advisor: Wu, S. Felix
Commitee: Chuah, Chen-Nee, Rowe, Jeff
School: University of California, Davis
Department: Computer Science
School Location: United States -- California
Source: DAI-B 73/07(E), Dissertation Abstracts International
Subjects: Computer science
Keywords: Crawling, Preferential attachment, Recommender systems, Social networking, User engagement, User lifetime
Publication Number: 3499459
ISBN: 978-1-267-23906-8
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy