Social Media Interaction-Based Mental Health Analysis with a Chat-Bot User Interface

In recent times, social media has played a major role in shaping the state of mind of young adults. This means that the content that is interacted with online may subconsciously have a significant effect on one’s mental health. This paper presents a novel approach to detect the mental health state of a user by analyzing their online activity over a period of time and generating a report indicating the same. The user is then allowed to ask any clarifying questions on the generated report, along with general queries on mental health via a chat-bot interface. In order to classify the mental state of the user based on the kind of content posted or interacted with by them, their tweets are scraped and feature vectors of the same are generated. Supervised machine learning algorithms like Support Vector Machines (SVM) and neural network-based models like Long Short-Term Memory (LSTM) are compared for their performance on prediction. A transfer learning approach is also attempted and gives promising results in predicting the classes of the tweets. Natural Language Processing techniques such as question similarity and extractive summarization are utilized in building the chat-bot framework.