A big issue with the semantic web is ontologies - the use of consistent definitions to concepts. For those that don’t understand what I’m talking about - essentially, the next evolution of the web is about making content readable by not just humans but also machines. However for a machine to understand something it reads, it needs consistent definitions. Human’s for example, are intelligent - they understand that the word “friend” is also related to the word “acquaintance”, but a computer would treat them to mean two different things. Or do they?
Just casually looking at some of my web analytics, I noticed some people landed on my site by doing a google search for how many acquaintances do people have, which took them to a popular posting of mine about how many friends people have on facebook. I’ve had a lot of visitors because of this posting, and its been an interesting case study for me on how search engines work. However today was something different from other times: I found the word acquaintance weird. I know I didn’t use that word in my posting - and when I went to the Google cache I realised something interesting: because someone linked to me using that word, the search engine replaced the word ‘friend’ with ‘acquaintances’.
Google’s linking mechanism is one powerful ontology generator.
I’m going to gather my thoughts on this issue of semantics. I spent quite a bit of time studying semantics while at uni, in particular private language theory. There’s probably a phd thesis waiting to be written about private language theory and the semantic web. I’ll gather my thoughts and blog about it if I can.
very interesting discovery.
are there any open hierarchical ontologies ? i am considering using the ontology on the open directory project.
i have used yahoo term extraction api, but the keywords generated is very dismal, especially for languages other than english.