I am about to evaluate (mostly) opensource reasoners like pellet, fact++, hermit, ... . So until now I've found some papers benchmarking reasoners (there aren't a lot), been on their respective homepages and so on, but still I find it hard to find arguments why one reasoner is better than another (considering for example the ones named above). Can anyone help me find a reasonable :-) approach to choosing the adequate reasoner? The application is not yet elaborated (the ontology language is not picked yet), so something like a list of general features, strength, weaknesses and the like would be great. Anyone having experiences with soundness, completeness, performance, ... of different reasoners?