Article Preview
Top1. Introduction
Although Internet-based video content and social media have proliferated with the development of IT technologies, and television (TV)-viewing behaviors have changed with the popularization of mobile devices and services, such as YouTube, Netflix, and other video-on-demand (VOD) applications, watching TV remains a major leisure activity. The 2017 American Time Use Survey stated that TV viewing accounted for the majority of leisure time (2.8 h per day), accounting for just over half of all leisure time, on average (Source: American Time Use Survey provided in https://www.bls.gov/tus). In addition, although the time spent VOD viewing on TV screens has consistently increased, live and scheduled linear TV content still represents almost 80% of the total viewing time of viewers (Source: ERICSSON CONSUMERLAB TV and MEDIA 2017). As changes continue to occur, understanding TV viewing behaviors remains important.
The development of technology has also changed the methodology of investigating TV viewing patterns. Studies on TV viewing behavior have traditionally conducted using surveys and attempted to reveal the relationships between demographic characteristics and TV viewing behaviors (Anderson et al., 1986), how television viewers selected programs (Brosius et al., 1992), and commercial avoidance behaviors or the effects of TV advertising (Moriarty and Everett, 1994; L.C.M. et al., 2003). However, it is known that responses to questionnaires or self-reporting diaries can be highly unreliable, and surveys of this type require considerable time and cost to conduct, since they must involve a large number of respondents for reliable analysis.
Recently, Internet Protocol TV (IPTV), cable TV, and satellite TV providers have included log collection systems in set-top boxes to gather comprehensive information about their subscribers’ viewing activities. Log data are more accurate than survey data and self-reporting diaries and can provide detailed information regarding TV viewing behaviors.
Utilizing the rich information in the log data, studies to understand TV viewers’ behaviors in depth and detail have increased. Most studies on viewers’ behaviors and TV watching have analyzed channel-switching behaviors (Becker, 2016). Employing viewer behavior analysis can help increase an audience’s rate, serve as the basis for a recommendation system for channels or programs, and help profile viewers so that they can be provided personalized advertising and new product marketing (Abrahamsson and Nordmark, 2012; Iguchi et al., 2015; Yu et al., 2017). In addition, researchers have used the log data, especially for IPTV, to attempt to better understand TV viewing patterns and predict the next program, in order to mitigate zapping delay (Ramos et al., 2011; Ferro et al., 2016; Basicevic et al., 2018).