Mining Health-Related Issues in Consumer Product Reviews by Using Scalable Text Analytics.

Area of Science:

• Computational linguistics
• Public health informatics
• Consumer behavior analysis

Background:

• Digitized life activities create vast data for population health insights.
• Online product reviews are an underexplored source of health-related information.
• Systematic mining of health information from consumer reviews is feasible.

Purpose of the Study:

• To quantitatively and qualitatively analyze health issues in consumer product reviews.
• To develop a machine learning classifier for detecting health-related issues in reviews.
• To identify text analytics challenges and characteristics for future research.

Main Methods:

• Utilized natural language processing and machine learning tools.
• Analyzed 1.3 million grocery product reviews.
• Performed quantitative and qualitative analysis of identified health information.

Main Results:

• Successfully mined health-related information from a large dataset of online reviews.
• Developed a machine learning classifier capable of detecting health issues in text.
• Characterized the types of health issues and challenges in consumer review data.

Conclusions:

• Online product reviews are a viable, albeit scarce, data source for public health surveillance.
• Machine learning and NLP are effective tools for extracting health insights from unstructured consumer data.
• Further research in text analytics is needed to refine health information extraction from online platforms.