Extracting features with medical sentiment lexicon and position encoding for drug reviews

Journal Publication ResearchOnline@JCU
Liu, Sisi;Lee, Ickjai
Abstract

Medical sentiment analysis refers to the extraction of sentiments or emotions from documents retrieved from healthcare sources, such as public forums and drug review websites. Previous studies prove that sentiment analysis for clinical documents has the potential for assisting patients with information for self assessing treatments, providing health professionals with more insights into patients’ health conditions, or even managing relations between patients and doctors. Nevertheless, the lack of data used for empirical experiments in previous research indicates that there are strong needs for a systematic framework in order to identify medical field specific sentiments. We propose a new feature extraction approach utilising position embeddings to generate a medical domain enhanced sentiment lexicon with position encoding representation for drug review sentiment analysis. Experiments on different feature extraction methods using two types of sentiment lexicons with various machine learning classifiers, support the superior performance of sentiment classification with position encoding incorporated medical sentiment lexicon for drug review datasets.

Journal

Health Information Science and Systems

Publication Name

N/A

Volume

7

ISBN/ISSN

2047-2501

Edition

N/A

Issue

11

Pages Count

10

Location

N/A

Publisher

Springer International Publishing

Publisher Url

N/A

Publisher Location

N/A

Publish Date

N/A

Url

N/A

Date

N/A

EISSN

N/A

DOI

10.1007/s13755-019-0072-6