A novel approach to track public emotions related to epidemics in multilingual data

Abstract

Emergence of new epidemic and re-appearance of older diseases causes great impact towards public health. Surveys based techniques which are costly and time-consuming are the most popular methods to measure information related to public health and used in decision making. Early monitoring of these epidemics helps in rapid decision making. Social media platforms provide rich source of information related to public health in forms of blogs, tweets, public posts etc., but these data is in unstructured form contains multiple languages words. This research focused on developing an automatic system for detecting public emotions related to epidemics in multilingual unstructured data to gain deeper understanding of public emotions and health related information. This approach gives timely information related to epidemics, corresponding symptoms, prevention techniques and awareness, which can help government and health agencies for rapid decision making. Experimental analysis of data set provides results that significantly beat the baseline term counting methods used for sentiment analysis.