So you are just guessing. I looked into the link you provided and also into wikipedia article on concept search that is linked there, and I do not see any evidence that the search engines at the major microstock sites work that way, except that they may address issues like various forms of the same word (strike vs. striking was given as an example) and sometimes minor misspellings, but the issues with polysemy and synonymy.
I wonder, did you expect to find in the link "Shutterstock uses algorythm XYZ to serve your files (just like ribtoks said)"?
You see, I'm not here to prove something to you. I shared my (engineering) insight, if you specifically do not agree with it, I have nothing against it. Luckily it does not mean it's wrong.