Opinion Mining for Measuring the Social Perception of Infectious Diseases. An Infodemiology Approach

Prior to the digital era, knowing the perception of society towards the health-system was done through face-to-face questionnaires and interviews. With this knowledge, governments and public organizations have designed effective action plans in order to improve our quality of life. Nowadays, as a result of the irruption of computer networks, it is possible to reach a higher number of people with a minor cost and perform automatic analysis of the collected data. Infodemiology is the research discipline oriented to the study of health information on the Internet. In this work, we explore the reliability of Opinion Mining to measure the subjective perception of people towards infectious diseases during times of high risk of contagion. In short, linguistic characteristics, among other relevant data, were extracted from tweets written in the Spanish Language by the end of 2017 in Ecuador. The built model contains the most relevant linguistics characteristics related to determine positive and negative pieces of text regarding infectious diseases. In addition, the corpus used in this analysis has been published for other researchers to use it in future experiments in this area. The results showed Support Vector Machines achieved the best results with a precision of 86.5%.