Sentiment Analysis on Tweets about Diabetes: An Aspect-Level Approach.

Area of Science:

• Computational Linguistics
• Health Informatics
• Natural Language Processing

Background:

• Existing sentiment analysis methods in healthcare lack focus on the specific domain of diabetes.
• A gap exists in analyzing sentiment at the aspect level within documents like reviews, news, and social media posts.

Purpose of the Study:

• To propose and evaluate an aspect-level sentiment analysis method tailored for the diabetes domain.
• To leverage ontologies and N-gram techniques for precise sentiment orientation of specific aspects.

Main Methods:

• Developed an ontology-based aspect-level sentiment analysis approach for diabetes-related text.
• Utilized N-gram methods (N-gram after, N-gram before, N-gram around) to capture contextual sentiment.
• Collected and manually labeled a corpus of diabetes-related tweets for evaluation.

Main Results:

• The 'N-gram around' method demonstrated superior performance in sentiment analysis.
• Achieved a precision of 81.93%, recall of 81.13%, and an F-measure of 81.24% using the 'N-gram around' approach.

Conclusions:

• The proposed ontology-based aspect-level sentiment analysis method is effective for the diabetes domain.
• The 'N-gram around' technique provides a robust way to determine sentiment orientation for specific aspects in social media data.