Towards Accurate Conflict Detection in a VCS for Model Artifacts: A Comparison of Two Semantically Enhanced Approaches