HealthMine: A Tool for Social Media Text Mining in Health

Social media has become a major source of information in recent years, with millions of posts every minute if not seconds. Containing information on various topics like health, politics and sports, one cannot deny that social media has become a good leverage for the field of data analytics. The objective of this work was to apply techniques of text mining, data analytics, and machine learning to Implement a web application, HealthMine, which extracts and classifies relevant data collected from social media platforms, namely Twitter and MedHelp, following a health-related user query. This would spare the user the burden of filtering out irrelevant information and focusing more on what is relevant. However, since content on social media is user-generated, the reliability is dubitable and the advice of a certified medical practitioner is always recommended. The main purpose of the tool is to allow users to share experiences. The tool was evaluated using 1400 tweets and 1800 MedHelp posts and it was found that Naive Bayes Classifier yielded the best accuracy among other classifiers, with an accuracy of 86.3 and 76.6 for Twitter and Medhelp respectively.