Healthcare information extraction from internet forums

• Main Author: Chen, Timothy Yang
• Other Authors: Sun Aixin
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/62584
• Institution: Nanyang Technological University

Despite an increasing number of people in Singapore seeking help for depression, there is still a large number who are not coming forward. With Internet forums being an increasingly popular medium for people to post their thoughts and feelings, it represents a good source to understand more about the topics that people who are sad care about, or to even identify users who are at risk. The identification of such topics or users could help in the treatment or diagnosis of depression. In this study, topic modeling is performed on several groups of posts to identify the most common topics that are found within posts that contain depressive language. This study also seeks to give a rough definition on what is deemed to be depressive language by continuous refinement of the words through iterations of topic modeling. The specificity of the words were deemed to constitute depressive language was then calculated using clarity scores. Additionally, the study also looks at the linguistic traits of posts with depressive language. Relationships were found to be a topic that was frequently found amongst the posts with depressive language, but not in posts without depressive language. This suggests that it is the topic that matters the most for Singaporean forum users who use depressive language.