ARCHIVES

Abstract: During this work we tend to examine on-line user reviews inside the pharmaceutical field. This data are often leveraged to get valuable insights victimisation data processing approaches like sentiment analysis. on-line user reviews during this domain contain data associated with multiple aspects like effectiveness of medicine and aspect effects, that create automatic analysis terribly attention-grabbing but conjointly difficult. However, analyzing sentiments regarding the various aspects of drug reviews will offer valuable insights, help with higher cognitive process and improve observation public health by revealing collective expertise. In this preliminary work we tend to perform multiple tasks over drug reviews with information obtained by creeping on-line pharmaceutical review sites. we tend to 1st perform sentiment analysis to predict the emotions concerning overall satisfaction, aspect effects and effectiveness of user reviews on specific medication. to satisfy the challenge of lacking annotated information we tend to more investigate the interchangeability of trained classification models among domains, i.e. conditions, and information sources. In this work we tend to show that transfer learning approaches are often used to exploit similarities across domains and may be a promising approach for cross-domain sentiment analysis.