Sentiment mining in twitter for early depression detection / Najihah Salsabila Ishak

Depression is a severe and pervasive threat to public health. These people like to express their thought, opinion and suggestion using social media network. Twitter is a popular microblogging site for users to post status updates (tweets). These tweets often reflect views on social issues, including psychological. Sentiment Analysis refers to natural language processing and text mining approaches to classify thoughts or sentiments from the tweet. Machine learning is an implementation of artificial intelligence (Al) that allows systems to learn and build on knowledge without being directly programmed automatically. This paper applies sentiment analysis, text mining, and machine learning to psychology to identify depression in Twitter user. The usefulness of using the user's tweet to measure depression studies using a literature review. The utility of current Python sentiment tools to a set of vocabulary used in microblogging is determined. The use of linguistic features to detect the sentiment in Twitter tweets are explored. A classifier model is developed using Naive Bayes characteristics. A comparison between built-in Scikit Learn Naive Bayes algorithm, and the scratch Naive Bayes algorithm is used to measure its effectiveness in terms of accuracy. At the end of this project, a prototype that can classify tweet is developed and used to monitor the tweets' sentiment probability.