Semantic web, semantic search, semantic this, semantic that – it’s enough to make a linguist or philosopher sigh with despair when the term is bandied about like this. Within specific domains such as philosophy or linguistics, there is a pretty clear and widely accepted understanding of semantics: the study of meaning. John Lyons (in 1997) even wrote two volumes about it.
But the use of “semantic” in IT is much less constrained. The populace as a whole uses the term as one of those words that conveniently means something vaguely good, without worrying too much about how you get there – you know the category, words such as “sustainable”, “community” “fair-trade” – words that sound good, even if on examination no two people might agree just what they mean. Unfortunately, the terms are usually never examined in more detail.
Even people who know a thing or two in this area identify vast areas of meaning within the word “semantic”. Seth Grimes, for example, in a 2010 article in Information Week, identifies no fewer than eleven different kinds of “semantic” search. They might all be called semantic, but are they? In several pf his cases, I’m convinced that “semantic” just means “clever”.
One thing is pretty clear: in the term “semantic linking”, there is no genuine meaning analysis going on at all. For confirmation, see Antoniou and van Harmelen (2008), who state that the goal of the semantic web is to provide “machine processable” web content, (not “machine understandable”). Once you realise the limitations of the term “semantic”, the remainder becomes much simpler.