COPOS: Corpus Of Patient Opinions in Spanish. Application of Sentiment Analysis Techniques

Flor Miriam Plaza del Arco, M. Teresa Martín Valdivia, Salud María Jiménez Zafra, M. Dolores Molina González, Eugenio Martínez Cámara


Every day more users are interested in the opinion that other patients have about a physician or about health topics in general. According to a study in 2015, 62% of Spanish people access the Internet in order to be informed about topics related to health. This paper is focused on Spanish Sentiment Analysis in the medical domain. Although Sentiment Analysis has been studied for different domains, health issues have hardly been examined in Opinion Mining and even less with Spanish comments or opinions. Thus we have generated a corpus by crawling the website Masquemedicos with Spanish opinions about medical entities written by patients. We present this new resource, called COPOS (Corpus Of Patient Opinions in Spanish). To the best of our knowledge, this is the first attempt to deal with Spanish opinions written by patients about medical attention. In order to demonstrate the validity of the corpus presented, we have also carried out different experiments with the main methodologies applied in polarity classification (Semantic Orientation and Machine Learning). The results obtained encourage us to continue analysing and researching Opinion Mining in the medical domain.

Texto completo: