External data for news algorithms

Master Thesis Research Project: January to June, 2016

Enhancing cross data understanding for creating news recommendations. Traditionally user behaviour is analysed on the news site in question – but what if you could analyse online behaviour from outside news and apply that information to news recommendation?

Lasse Daniel Sander Jensen & Nicolai Colding

IT & Communication Studies, Aarhus University

Project partners:


This thesis seeks to explore how different data types can be combined to create a better basis for recommending relevant news content. Instead of simply analysing a user’s behaviour on a single news website or app (news brand) as has been the standard for years, this Master Thesis aims to include many different types of data sources, such as activity on Facebook, Twitter, web behaviour, location data and even shopping preferences, mood (data from fitness trackers, smartwatches etc) and many more. Both personal and general data will be included from either a static or dynamic point of view. Furthermore the thesis will explore how the data can be processed to generate knowledge, paving the way for the creation of innovative and highly personal end-user products.

Below the surface

To illustrate the main topic of our thesis, we like to think of it as an iceberg. Above the surface we find the layer of user experiences and underneath this a big machine is presented. A product of the iceberg is for instance the well known Netflix, where an algorithm beneath the surface presents relevant content to the user. We expect that the same kind of algorithms exploring patterns in both personal and general data from different data sources can have a great impact in the world of news media.

#Allprojects #Previousprojects