User Behavior Analysis And System Design In New Media Systems

Due to the development of Information Communication Technology (ICT), new media systems are becoming increasingly prevalent. These new media systems support a great number of users with complex interactions. Massive data are generated from these systems which contain rich information. Besides that, it is getting hard to regulate users’behaviors in new media systems. These new characteristics bring more challenges for user behavior analysis and system design. Traditional methods are not able to thoroughly analyze new media behaviors. New media systems are not flexible enough because the design approaches are not built upon deep understandings of complex user interactions, which may cause many online and offline social problems. Adapted from socio-technical theory, this dissertation proposes socio-technical methods for behavior analysis and new media system design in three typical new media systems (i.e. online gaming, online TV watching and microblogging sites). The main work and contributions are listed in the following four aspects:(1) A tightly-coupled socio-technical framework is proposed for user behavior analysis and system design. This framework contains a set of methods for data capture, data analysis and system design. For user behavior analysis, technically, data mining algorithms are designed to analyze system logs. Socially, empirical studies are applied to analyze user behaviors. For system design, social findings are integrated into system design process with social science theories to support contingent interactions. Besides combining social aspect and technical aspect, this framework combines analysis and design processes together to improve system flexibility.(2) Technical-to-social behavior analysis methods are proposed and used to analyze watching and gaming behaviors on an online arcade gaming platform in the perspective of infrastructure. To investigate popular watching phenomena and gaming experience on ArcLive, a hot Chinese online arcade game platform with over 6 million subscribers, this study starts from a technical exploration. Grounded on the system log analysis, game watching is temporally synchronized and spatially concentrated. A community detection algorithm is applied to further explore the new community structures. Then an empirical study is conducted and the findings show that user experiences are highly related to technological, geographical and physical infrastructures, which also help generate new community structures. This study innovatively demonstrates socio-technical phenomena on ArcLive with an infrastructural perspective and finds how the community develops its own norms and cultures, and examines the relationships between system technologies and social interactions, all of which can guide system design for online gaming platforms.(3) Social-to-technical behavior analysis methods are proposed to analyze digitality and materiality related TV watching behaviors and patterns on a new media TV platform. On PPTV, one of the popular online TV watching platforms in China, qualitative methods are used to examine how the digital and material properties of new media shapes different watching patterns in terms of background TV, deep watching, social TV, etc. Then a watching preference mining algorithm is designed to verify the deep watching pattern from social findings. This study demonstrates the complementary effects of digitality and materiality of new media TV, and provides valuable insights describing practices, expressions and socializations, which can inspire the design of new media TV platforms.(4) A rumor transmission model is constructed to quantitatively analyze rumor transmission mechanisms on microblogging platforms. A weibo analysis system is implemented for rumor transmission problems. On Sina Weibo, one of the most popular microblogging platforms in China, a study is conducted to demonstrate how event contextual factors (i.e., topic, emergency degree and influence coverage) and non-text clues (e.g., identity verification, additional images, devices and etc.) influence online rumor transmission. Based on the findings of the rumor transmission mechanism, a social translucence approach is applied in Sina Weibo Analysis Plugin, which visualizes weibo information to help people judge the credibility of certain weibo posts, and accordingly promote their accountabilities.