Gender Classification Using Sentiment Analysis and Deep Learning in a Health Web Forumopen access
- Authors
- Park, Sunghee; Woo, Jiyoung
- Issue Date
- 25-Mar-2019
- Publisher
- MDPI
- Keywords
- sentiment analysis; gender classification; machine learning; deep learning; medical web forum
- Citation
- Applied Sciences-basel, v.9, no.6
- Journal Title
- Applied Sciences-basel
- Volume
- 9
- Number
- 6
- URI
- https://scholarworks.bwise.kr/sch/handle/2021.sw.sch/4642
- DOI
- 10.3390/app9061249
- ISSN
- 2076-3417
- Abstract
- Sentiment analysis is the most common text classification tool that analyzes incoming messages and tells whether the underlying sentiment is positive, negative, or neutral. We can use this technique to understand people by gender, especially people who are suffering from a sensitive disease. People use health-related web forums to easily access health information written by and for non-experts and also to get comfort from people who are in a similar situation. The government operates medical web forums to provide medical information, manage patients' needs and feelings, and boost information-sharing among patients. If we can classify people's emotional or information needs by gender, age, or location, it is possible to establish a detailed health policy specialized into patient segments. However, people with sensitive illness such as AIDS tend to hide their information. Especially, in the case of sexually transmitted AIDS, we can detect problems and needs according to gender. In this work, we present a gender detection model using sentiment analysis and machine learning including deep learning. Through the experiment, we found that sentiment features generate low accuracy. However, senti-words give better results with SVM. Overall, traditional machine learning algorithms have a high misclassification rate for the female category. The deep learning algorithm overcomes this drawback with over 90% accuracy.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - SCH Media Labs > Department of Big Data Engineering > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.