Exploring correlation between large-scale social network service posts and box office grosses

초록

The film industry represents a highly profitable business since the global box office is continuously increasing in terms of the economic scale. However, only a few movies were financially successful in the last decade. The film industry is a risky, uncertain, and complicated business, and thus several studies suggested movie prediction models to predict financial success of a movie. The increasing use of social network services, such as Twitter, provide individuals to obtain easier access to customers’ feedback on a movie in which it constitutes a valuable source to predict the financial success of the movie via natural language processing (NLP) techniques. This study presents a methodology to extract users’ sentiments from movie related tweets to explore relationships between massive electronic word-of-mouth (eWOM) data sets and box-office grosses. The study hypothesizes that correlations exist between eWOM and movie box office grosses. This study analyzes each correlation between the number of tweets and box office gross and sentiment of the tweets and box office gross. The case study tests the hypothesis with 42,590 movie related tweets. To automatically process massive data, the research model involves performing text mining techniques and statistical verification models. Therefore, the model extracts 42,590 movie tweets and the sentiments from 1,237,454,287 tweets collected from March 2011 to August 2012 in North America. This case study reveals that the tweets and their sentiments extracted by the method are correlated with the box office gross.