Google Semantic Web 2006/05/06
Google Operating System (a Google watch-blog) is reporting on Google’s recent testing of some word categories and other methods of prompting users to refine their search terms as part of their effort to build something called Google Semantic Web.
This is something I’ve been looking for out of Google for quite a while now, so I’m glad to see they’re beginning to experiment with it more publicly.
If you ask an average Google search user, their biggest complaint isn’t that they can’t find what they’re looking for on Google - it’s that in order to do so, they often have to rephrase their query several times before the results they were looking for make their way to the top of the list -
In other words, the traditional “search problem” has been reversed.
It is no longer Google’s ability to index the web that is limiting the relevance of the search results - it’s the inherent quality (or lack thereof) of the user’s query.
The problem for Google (and any other search company) is that even if their AI is the best in the world - if Google’s search box was just as intuitive as another human being - even then, could it ever return exactly what the user had in mind with only vague, one word queries to go on?
I’d say not. Google’s search box could benefit from improved semantics, certainly, but only in order to interpret complete queries more like a human might. Despite what GoogleOS hints at, no amount of complex algorithms are actually going to enable Google to flat out read your mind.
Some gentle prompting for the user to refine their query is a fairly elegant solution, in my opinion.
It’s not as good as a Star Trek style conversational voice interface might be someday down the line, but certainly a step in the right direction, eh?