This YouTube video examines the question ” what is the future of semantic search“?  Some people describe semantic search as a way in which website creators or publishers specify relationships between words and concepts.  Whilst the basis of this was full of promise, in reality it turns out that the framework is really rather brittle.  Instead it may be more accurately described as a set of statistical data about the relevance of language and concepts which Google learns by comparing, for example, different language translations about similar information. 
This video points to the probability that the Google algorithm will ultimately determine what the user ‘intends’ to ask rather than simply look at different ways of expressing it.  It’s this interpretation that makes the video interesting.