Designing a social-broadcasting-based business intelligence system

Abstract

The rise of social media has fundamentally changed the way information is produced, disseminated, and consumed in the digital age, which has profound economic and business effects. Among many different types of social media, social broadcasting networks such as Twitter in the U.S. and “Weibo” in China are particularly interesting from a business perspective. In the case of Twitter, the huge amounts of real-time data with extremely rich text, along with valuable structural information, makes Twitter a great platform to build Business Intelligence (BI) systems. We propose a framework of social-broadcasting-based BI systems that utilizes real-time information extracted from these data with text mining techniques. To demonstrate this framework, we designed and implemented a Twitter-based BI system that forecasts movie box office revenues during the opening weekend and forecasts daily revenue after 4 weeks. We found that incorporating information from Twitter could reduce the Mean Absolute Percentage Error (MAPE) by 44% for the opening weekend and by 36% for total revenue. For daily revenue forecasting, including Twitter information into a baseline model could reduce forecasting errors by 17.5% on average. On the basis of these results, we conclude that social-broadcasting-based BI systems have great potential and should be explored by both researchers and practitioners.