Measure Your Filter Bubble

This master thesis introduces a first prototype of an agent-based model that allows people to check their risk exposure of living in a filter bubble based on their social media newsfeeds as well as the homogeneity of their social network. Filter bubbles first became popular in 2011 when Eli Pariser introduced the term. Even though, they really started to be talked of when Donald Trump was elected President of the United States of America. Filter bubbles are the results of today's online personalisation methods. Without a newsfeed in social media platforms, people are over flooded with posts. Based on the user history (posts, likes, shares etc.) a filtering process is responsible to show the user only the most relevant content. Latest political events showed that people were manipulated due to a filter bubble, and brought the topic into the spotlight of the media and research. Since today no tool is available for people to measure their risk exposure to a filter bubble, the goal of this thesis was to develop a first prototype to close this gap. Due to the characteristics of agent-based modelling this approach was evaluated as promising and used to build the model in NetLogo. This simulation needed an underlying model that detects the risk of a filter bubble and translates it to a risk score. This was also developed during this thesis. Due to the complexity of analysing the newsfeed of users and categorising posts, it was decided to simplify the developed prototype. The model knows only seven different post topics and so far, it was only run with data of generic scenarios. Even though, the analysed and evaluated results generated by the model seem realistic. Therefore, agent-based modelling is still recommended for future work in the area of measuring the risk exposure of a person to a filter bubble.