An Assessment of ML-based Sentiment Analysis for Intelligent Web Filtering

Abstract

Many people are increasingly using the Web both to accomplish tasks but also for entertainment. At the same time, for many people a significant fraction of their web-browsing time is spent on reading news, both from social media sources and traditional news outlets. However, often this comes with a mental price when facing negative news: research shows that doomscrolling can have a negative impact on your well-being. In this paper, we discuss the development and evaluation of an intelligent web filtering mechanism, in the form of a web browser extension (i.e., plugin). This mechanism aims at providing an intelligent assistant that filters out undesired content, aiming at improving the user's well-being. The effectiveness of this approach is assessed with an evaluation involving 50 participants. Our results show that our approach is acceptable by the participants and that there is indeed a need for such tools.