Identification of Ontologies of Prediabetes Using SVM Sentiment Analysis

Abstract

Sentiment analysis is considered as a classification task as it classifies the polarity of a text into positive or negative. Different methods of sentiment analysis can be applied for the health domain, especially for prediabetes domain which has not been completely explored yet. And there is a lack of approaches for analyzing positive and negative tweets separately to identify the positive and negative ontologies for modeling the features in a domain of interest. Here in my work, proposed domain and sub-domains are Health and Prediabetes, respectively. Prediabetes defines the condition of blood sugar levels that are higher than normal but not high enough to be diabetes like a pre-warning call for diabetes. The proposed methodology is the deployment of original ontology-based techniques toward a more efficient sentiment analysis of Twitter posts on prediabetes. As part of experimentation, sentiment analysis uses the SVM algorithm with term frequency as a feature extraction method to train and test a large and sub-data set of tweet text. Negative ontologies are constructed for a better understanding of the aspects identified through semantic annotations. The results of the classification method are evaluated using the performance metrics accuracy, precision, recall, and F-measure for effective evaluation of the proposed method.