Analytic approach of understanding crowd phenomena on the internet

Abstract

Quantification of collective human behavior and understanding the group characteristics in the Internet is important in user behavior studies since people tend to gather together and form groups due to their inherent nature. On the Internet, people are also often forming a group for a specific purpose such as i) an online group in games (e.g., MMORPGs) to experience various social interactions with other players or accomplish a difficult quest with teammates or ii) a swarm in peer-to-peer network to share a content to utilize a higher download rate with an availability. To this end, we studied the two most well-known major applications in the Internet that people are actively using with different purposes

i) MMORPGs and ii) BitTorrent. In this dissertation, we analyze the i) group activities of users in Aion, one of the largest MMORPGs, based on the records of the activities of 94,497 users and ii) crowd phenomena of BitTorrent. First, in a case study of Aion, we focus on (i) how social interactions within a group differ from the ones across groups, (ii) what makes a group rise, sustain, or fall, (iii) how group members join and leave a group, and (iv) what makes a group end. We first find that structural patterns of social interactions within a group are more likely to be close-knit and reciprocative than the ones across groups.We also observe that members in a rising group (i.e., the number of members increases) are more cohesive, and communicate with more evenly within the group than the ones in other groups. Our analysis further reveals that if a group is not cohesive, not actively communicating, or not evenly communicating among members, members of the group tend to leave. Second, we investigate what kinds of crowd phenomena of content exist and why different patterns of crowd phenomena appears and how we can exploit content crowd phenomena considering the content category, publisher, and population of content in BitTorrent. To this end,We conduct comprehensive measurements on content locality in one of the largest BitTorrent portals: The Pirate Bay. In particular, we focus on (i) how content is consumed from spatial and temporal perspectives, (ii) what makes content be consumed with disparity in spatial and temporal domains, and (iii) how we can exploit the content locality. We find that content consumption in real swarms is 4.56 times and 1.46 times skewed in spatial (country) and temporal (time) domains, respectively. We observe that a cultural factor (e.g., language) mainly affects spatial locality of content. Not only the time-sensitivity of content but also the publishing purpose affects temporal locality of content.We reveal that spatial locality of content iii rarely changes on a daily basis (microscopic level), but there is notably spatial spread of content consumption over the years (macroscopic level). Based on the observation, we conduct simulations to show that bundling and caching can exploit the content locality.