Sentiment Manipulation in Online Platforms: An Analysis of Movie Tweets

• 成果类型: Article
• 署名作者: Lee, Shun-Yang; Qiu, Liangfei; Whinston, Andrew
• 署名单位: University of Connecticut; State University System of Florida; University of Florida; University of Texas System; University of Texas Austin
• 刊物名称: PRODUCTION AND OPERATIONS MANAGEMENT
• ISSN/ISSBN: 1059-1478
• DOI: 10.1111/poms.12805
• 发表日期: 2018
• 页码: 393-416
• 关键词: BOX-OFFICE PERFORMANCE; instrumental variables; Social media; reviews; IMPACT; sales; user; COMPETITION; management; reputation

摘要

Online platforms are prone to abuse and manipulations from strategic parties. For example, social media and review websites suffer from sentiment manipulations, manifested in the form of opinion spam and fake reviews. The consequence of such manipulations is the deterioration of information quality as well as loss in consumer welfare. We study one of movie studios' operation activities, sentiment manipulation, in the context of movie tweets. Using the movie release and movie studios' earning announcement dates as sources of exogenous shocks, we find that both the average Twitter sentiment and the proportion of highly positive tweets exhibit a significant drop on the movie's release day or movie studios' earnings announcement day. In addition, independent productions and low budget movies tend to experience a larger drop than major studio productions and high budget movies. To examine the effect of competition on firm manipulation, we construct a movie competition measure based on both the time and theme dimensions through topic modeling, and we find that a higher level of competition leads to a larger drop in Twitter sentiment. Overall, these observations suggest that firms might be actively managing online sentiment in a strategic manner. Our study sheds light on the reliability of sentiment analysis and contributes to our understanding of potential strategic manipulation in the operation of movie studios.