Application to Sentiment Analysis

This chapter illustrates the building and expansion of WordNet for Medical Events (WME) and evaluate its performance. WME has been developed for medical opinion mining and can be used as a standalone medical lexicon. ConceptNet has been used to improve the graphical representation of the underlying architecture in WME. Two methods have been proposed and incorporated to improve the overall performance of the lexicon. First method adds two new features to the existing WME namely affinity and gravity score. To evaluate the new structure, machine learning techniques and linguistic approaches have been incorporated. Finally, the chapter proposes a novel fusion of computational creativity and machine learning.