No criteria, but I believe google academics is not only simple but use good data mining techniques in the search of patterns and ontologies. Of course we can compare the google results with other search machines in order to form patterns and find divergences! Regards, Farley.
I was going to say, don't forget Linked Open Vocabularies, but site is not working today. http://labs.mondeca.com/dataset/lov/index.html . Anyone know what's up?
However, in my opinion it is rather unsatisfying that semantic web technology, which aims to simplify the access to information, is itself documented and made available in such a scattered and inconsistent way like it is today: many outdated or unreachable sites (see answers above, and also last part of this blog post: https://cknoll.github.io/semantic-python-subjective-notes-en.html ), little tool support (apart from protoge and owl-api).
From my perspective a desirable situation would be like for "numerical machine learning" (neural networks etc.): lots of up-to-date documentation and tools (in various languages).