Intelligent Widgets Showing Recommendations for News Articles

Abstract

This thesis addresses the use of intelligent widgets and user-based recommender systems for news articles to improve the end-user experience by providing recommendations that retain user attention effectively. In contrast to traditional news outlets, be it physical or digital news- papers, we explore ways to leverage social media UX concepts as they dominate the space of content consumption nowadays, specifically by the introduction of what we call "Intelligent Widgets" which are inspired by the popular story UI on social media and display personalized recommendations. To go along with this, we explore various types of recommender systems and mainly focus on collaborative filtering algorithms like Alternating Least Squares or Neural Collaborative Filtering. The goal is the development and deployment of a prototype of a digital news website incorporating "Intelligent Widgets" which we can use to collect user feedback about the user experience of interacting with news articles in this novel way. In the user study we also draw comparisons to the way personalized recommendations are usually displayed on online newspapers as a list. The results show a potential preference for interacting with Intelligent Widgets of around 60% over the conventional methods. The user-feedback also offers opportunities to further investigate and improve upon the concept of Intelligent Widgets.