Opinion Mining in Online Social Media for Public Health Campaigns

Tobacco smoking remains the leading cause of preventable death and disease in the United States. Since 2013, U.S. Centers for Disease Control and Prevention (CDC) has launched public health campaigns called "Tips" to build public awareness of the immediate health damage caused by smoking and encourage smokers to quit. This paper focus on evaluating audience reaction about anti-smoking campaign, Tips on online social network, which is modeled as an opinion mining problem. To solve this problem, we propose the model Community-based Sentiment Analysis Method (ComSenti), which incorporates social relation among users and sentiment signals from terms into a unified framework. Extensive experiments on Tips social data of year 2013 and 2015 corroborate ComSenti's effectiveness. Based on ComSenti, we also find that Tips 2015, the anti-vaping campaign received more opponent posts since the harm of e-cigarette is controversial.