What is in a recommendation? The case of the bX Article Recommender

Abstract

This thesis explores the bX Article Recommender, a recommender system for research papers, developed by Ex Libris. It views the system as embedded in a historical and social context, and studies it from both a technical and epistemological perspective. Its technical components are studied by asking how it harvests data and generates recommendations on the basis of this data. Furthermore it explores implicit epistemological statements contained within the system. For example the system reflects and instrumentalises a certain interpretation of the phenomena ‘relatedness’. The thesis demonstrates how the bX Article Recommender’s output is influenced by human impact and decisions. When designing recommender systems many choices have to be made, choices with great implications for the produced recommendations.