Natural-Language-Processing-to-Understand-Peer-Reactions-to-Expressions-of-Stress

To understand how to positively reframe or react to expressions of negative feelings on social media.

Problem Statement:

Social media platforms are now the place where many people are expressing their feelings and their views on a given topic. Often, these feelings are negative and reflect the impact of stressful situations or contexts. We are working on a task called positive reframing in which we seek to identify (1) positive ways to reframe a negative point of view; and (2) peer reactions to negative feelings that receive endorsement (and thus are likely to have a positive impact).

Project Mission:

To understand how to positively reframe or react to expressions of negative feelings on social media. ##Project objective: The project has three main objectives. First, to identify peer reactions to negative expressions of stress; that is, given a statement on social media labeled as “stressed” by its author, we want to identify the reactions that peers have toward this statement, often to provide emotional support. Second, we want to analyze and understand the composition of such positive reactions, and especially the ones that receive endorsement (eg, likes) from the author of the original stressed statement. Finally, if time permits, we want to identify ways in which these pairs of stressed statements - positive reactions can be used toward the development of chat bots for positive reframing/feedback.

Project boundary

As we will not have access to twitter, we only experiment with the data, and our project will not have a direct impact on the users. We plan to find ways to reframe or reply positively to a statement expressing negative stress.

We have created a backend to run it do the following

go to app directory

cd app
sudo uvicorn app:app --reload