Abstract:
"
In these days’ social media like twitter became widespread to all generations. People
tend to connect and share their emotions, feelings, pursuits etc. on social media more
frequently. Depression is a been a major issue to all ages of people and if it’s not treated
immediately it may lead to make fatal decisions in life. Specifically, young generation
is becoming a main victim for depression. In this virtual era people are very comfortable
on sharing their emotions on social media rather than to a person this is because social
media has because a part of our lifestyle. Sentiment analysis has been taken into many
changes all over the years and have been user and various fields. Detecting depressive
tweets on twitter is a way of detecting depression. By performing a sentiment analysis
in twitter data on a specific user’s twitter account we could find out if he or she is
showing any signs of depression. This is the part where this research paper focus on.
Firstly, the tweets from the users account are fetched and the prediction of happens.
The prediction of depression is performed through a recurrent neural network Bi directional short-term memory (BLSTM). In this research paper it talks about fetching
live twitter data from a user account and Vectorizing the tweets and feeding it into the
neural model using GloVe embedding and detection of depression using through
BLSTM. Much research have been taken forwarded by researches, but they use the
already scrapped data or data from open source. In this project tweets will be fetched
according to the provided username and will be stored as a CSV and the prediction will
be performed with GloVe and BLSTM"
