Towards better TV viewing rates: exploiting crowd's media life logs over Twitter for TV rating

TV viewing rate is a critical indicator regarding the popularity of programs, eventually influencing the revenue of broadcast stations via advertisements. Currently, the major methodology for assessing viewing rates is the Nielsen TV rating, which depends on surveys using a small number of randomly selected representative groups, because of practical considerations such as cost and survey time. Hence, the method has been criticized for seeking to move the evaluation in a certain direction, depending on who participated in the survey. However, the present media environments are drastically changing our media consumption patterns; we can watch TV programs anywhere and anytime outside the home, where TV viewing was primarily focused in the past. Owing to the advance of the Internet and many social networking sites, surveying massive audiences' media lifestyles has become increasingly easier. In order to better evaluate TV viewing rates in the light of evolving TV lifestyles including the online video and mobile environments, we are proposing to estimate the public TV viewing rates by means of Twitter where we can monitor crowd voices relative to TV watching. With the utilization of wider range of people and broader crowds over Twitter, we can overcome the limitation of conventional TV ratings. In the experiment, we describe our exploratory surveys, conducted using a large amount of Twitter messages targeting TV programs in Japan.