Stress Level Assessment of an Individual using Neural Networks based on Tweets

Abstract

Stress in workplace is deteriorating both physical and mental health of an individual. It also makes the individual less productive and less efficient in work. It is necessary to assess the stress level before coping with it. There are so many methodologies that are used to detect stress. Inspired by psychological research stating that people feel more comfortable to express feelings on social media than through verbal communication, we are proposing a simple model to assess the stress level of an individual which can prevent psychological complications. In this paper, we propose a system similar to twitter that analyzes the stress level of an individual which can be viewed by the admin. Main emphasis is laid on stress level detection using Neural Networks and Semantic Similarity. Dataset, consisting of the stressed words is trained. Then, we find the similarity between tweet content and the trained dataset using Leacock and Chodorow’s (LCH) semantic similarity between words. Results show the stress level of each individual based on his/her tweet/s.