Sentiment Analysis on the News to Improve Mental Health

TL;DR

This paper presents a mobile app that uses multiple sentiment analysis models to curate uplifting news stories, aiming to improve mental health amid online negativity, with positive user feedback and reported mental health benefits.

Contribution

It introduces a novel pipeline combining four sentiment analysis models to filter news content for mental health support via a mobile application.

Findings

App received 4.9-star rating from 1,300 users

85% of users reported improved mental health

Successful implementation of multi-model sentiment analysis pipeline

Abstract

The popularization of the internet created a revitalized digital media. With monetization driven by clicks, journalists have reprioritized their content for the highly competitive atmosphere of online news. The resulting negativity bias is harmful and can lead to anxiety and mood disturbance. We utilized a pipeline of 4 sentiment analysis models trained on various datasets - using Sequential, LSTM, BERT, and SVM models. When combined, the application, a mobile app, solely displays uplifting and inspiring stories for users to read. Results have been successful - 1,300 users rate the app at 4.9 stars, and 85% report improved mental health by using it.