A Context-Aware Running Route Recommender Learning from User Histories Using Artificial Neural Networks

Sönke Knoch, Alexandra Theobalt, Andreas Emrich, Dirk Werth, Peter Loos

In: Giovanni Semeraro , Marco de Gemmis , Pasquale Lops , Markus Zanker (editor). 23. International Workshop on Database and Expert Systems Applications - DEXA 2012. International Workshop on Recommender Systems Meet Databases (RSmeetDB-2012) 23rd located at Database and Expert Systems Applications (DEXA) 2012 September 3-7 Vienna Austria IEEE Conference Publikation 2012.


So far, several websites exist where runners can request route information. Those systems are rather complex and lack a mobile-specific design. Thus, we propose a mobile running route recommender system (RRR) which supports the user while running or while planning the running route. The gathering and modeling of the route and its context/environment is discussed in respect of computational performance. A four dimensional plugin based ranking function is established that considers location-, time-, content-, and community-specific route features which cover all data types in our database. A conceptual model shows how the runner’s physical condition could be involved by predicting the heart rate for certain routes. Therefore, Artificial Neural Networks are chosen as data mining methodology to extend the existing recommender system.

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz