Written by Saad
Published under Technology
January 7, 2009
Google has started to display ‘so-called’ Semantic data for some of the search queries. For example a search for “Who is Barack Obama’s Wife?” in google gives me the following result:
Similarly if you do the opposite and search for – “Who is Michelle Obama’s Husband“, it does display the correct answer, however from a different source.
Why I think these aren’t regular Semantic Data?
I checked some of the source pages from where the results were derived from. I have not found a single semantic markup used in any of the pages that defines “Barack Obama” as Michelle Obama’s husband and so on. Thus the data that google is showing is either from an external page (which isn’t listed) that contains the structured data. If that is not the case then I guess Google has came up with a completely different revolutionary solution. Semantic web is far from being a reality at this moment, and it will take a long time before everyone starts to embrace it and do the ‘extra’ work so that the static information on the web can become ‘useful’ and can be processed by a computer.
So I believe Google has somehow taken a ‘short-cut’ approach to the problem with promising results. But of course the results are not 100% accurate. A search for “Yahoo CEO” gave me the following result -
Poor Jerry. See what I mean? Even on that referenced page there is no markup that defines Terry Semel as Yahoo’s CEO. So this leads me to believe that google is still playing with “keywords” – perhaps a bit more intelligently by using related phrases, position of the words, common verbs etc.
Google has done similar things before – for example in May 2008, Search Engine Roundtable reported that Google automatically highlights related terms in the SERP. So for example if you search for accredit, It automatically highlights words like – accreditation, accrediting etc.
Google can also answer direct questions like – “whats the capital of China“?
I couldn’t help noticing some of the comments that has been made about this by the ReadWriteWeb’s readers. Some of them are saying they have seen these before – well, I agree if we are talking about ‘what’s the capital of x’ type of queries. But some of the queries that Marshall Kirkpatrick of RWW came up with are completely new for me. But definitely these aren’t semantic or structured data. Because the results don’t reflect that. For instance a search for “Laura Bush Husband” gave me the following result:
A search for “Who is Justin Timberlake’s Girlfriend“, gave me the following results:
The source page, doesn’t even contain the keyword “girlfriend” anywhere.
Google is definitely onto something! Share your views in the comments!