The aim of this thesis is to address the problem of making automated recommendations of exercise plans for people involved in fitness and especially in training with weights, based on their personal preferences and existing training principles. Therefore, the following problem statement was formed:
How can a recommender-system application support fitness enthusiasts by producing automated and personalised weight-training exercise plans based on proven training principles?
Additionally to that, a few sub-questions were added, which lead to further investigation of the following topics: recommender techniques that can be used, fitness domain information that will be needed, ways to personalise the solution, applicable recommendation algorithms and how such an solution could be designed, implemented and tested.
The final solution prototype uses a hybrid recommendation system based on constraint based techniques and similarity heuristics to achieve the expected results.