W3C home > Mailing lists > Public > public-rww@w3.org > May 2012

Re: Google Knowledge Graph Experiment

From: Kingsley Idehen <kidehen@openlinksw.com>
Date: Fri, 18 May 2012 14:55:36 -0400
Message-ID: <4FB69B28.2000501@openlinksw.com>
To: Eric Franzon <eric@semanticweb.com>
CC: Ivan Herman <ivan@w3.org>, Aidan Hogan <aidan.hogan@deri.org>, "public-lod@w3.org" <public-lod@w3.org>, "semantic-web@w3.org" <semantic-web@w3.org>, "public-rww@w3.org" <public-rww@w3.org>
On 5/18/12 1:39 PM, Eric Franzon wrote:
> I also am still not getting to view GKG enrichments either. It truly 
> seems to be a gradual roll-out (as Google confirmed to us yesterday: 
> http://semanticweb.com/google-knowledge-graph-interview_b29172). I 
> have tried all three of my Google accounts, all of which have 
> completed profiles, but to no avail.

Remember, my *theory* is that this is an early example of a 
profile-driven-service roll-out. Nothing wrong with that, but Google 
made some disconcerting mistakes. Here's my sequence of events:

1. I open up the doc at: 
http://www.google.com/insidesearch/features/search/knowledge.html, 
following the service announcement
2. clicked on the "Give it a try" link
3. saw nothing
4. 24 hours later (yesterday EST) I stumbled across Paul's post (as per 
Twitter thread that ensued)
5. kicked off experiment to test my theory.

At step 1 above, nothing from Google indicated to me that if don't see 
anything it was due to staggered roll-out. In short, the disclaimer 
still doesn't exist on that page.

However you look at it, this is an algorithm gone wrong, and it does 
have implications.
>
> What the representative we interviewed indicated is that they are 
> rolling out slowly to account holders "over the next few days." THEN, 
> they will begin with general Google.com visitors. And then I suspect 
> they will (again gradually) start rolling it out across other global 
> Googles.

All vague commentary, the kind that is kinda scary and a little creepy. 
Again, I doubt it was intentional, but there are some realities bubbling 
up to the surface re. profiles and the Web etc..

>
> I was interested in Kingsley's experiment because it was based on the 
> theory that some measure of "social influence" might be at play in the 
> algorithm. While my ego might certainly wish to believe that I (and 
> semanticweb.com <http://semanticweb.com>) have such influence, I 
> acknowledge that Danny Sullivan, Kara Swisher, et al probably have 
> more in the broader tech world. ;-)

It isn't so simple. There are many other factors in play. There was a 
degree of profiling involved (nothing wrong with that per se.) that 
wasn't handled properly (IMHO).

>
> Of course, if Google would mention "Semantic Web" in their marketing 
> material or acknowledge any of the other work that has come before, 
> that would no doubt help with the "influence" of all of us on these 
> lists. But I digress... for more on /that/ perspective, you may wish 
> to see Sean Golliher's editorial which will be published at 
> SemanticWeb.com at the top of the next hour (2:00pm ET in the US).

Will take a look, its 2.55pm EST right now :-)

Kingsley
>
> Cheers,
> --Eric
>
>
>
> On Fri, May 18, 2012 at 9:45 AM, Ivan Herman <ivan@w3.org 
> <mailto:ivan@w3.org>> wrote:
>
>     Well... I tried this trick, but does not change a thing. Yes, the
>     search happens on www.google.com <http://www.google.com>, but I
>     presume it knows that I am, hm, considered as Dutch and sticks to
>     its guns: I see nothing of the GKG on my screen.
>
>     Ivan
>
>     On May 18, 2012, at 17:57 , Aidan Hogan wrote:
>
>     > All,
>     >
>     > On 18/05/2012 05:51, Eric Franzon wrote:
>     >> Actually, some modest testing has shown something other than
>     geography at play here. Earlier today, colleagues in London and
>     California were able to see GKG rich data visualizations, while
>     others in the US (myself included -- also in California) and
>     Europe could not.
>     >>
>     >> I spoke to a Google representative this afternoon who confirmed
>     the gradual roll-out, but would not (or could not) discuss the
>     algorithm. He did hint that people with Google accounts will be
>     first to see the enhancements.
>     >
>     > In my experience, the most reliable way to see GKG in action is:
>     >
>     > Visit http://google.com/ncr
>     > (the "/ncr" suffix stops you redirecting to a local version)
>     >
>     > Be logged in with a Google account.
>     >
>     > Try a typical entity search (e.g., "Galway").
>     >
>     > Cheers,
>     > Aidan
>
>
>     ----
>     Ivan Herman, W3C Semantic Web Activity Lead
>     Home: http://www.w3.org/People/Ivan/
>     mobile: +31-641044153 <tel:%2B31-641044153>
>     FOAF: http://www.ivan-herman.net/foaf.rdf
>
>
>
>
>
>
>
>
> -- 
> *Eric Axel Franzon*
> Vice President of Community
> SemanticWeb.com
> 6080 Center Dr., 6th Floor
> Los Angeles, CA 90045
>
> eric@semanticweb.com <mailto:eric@semanticweb.com>
> O: +1.323.856.1474
> C: +1.323.309.1601
>
> LinkedIn: http://linkedin.com/in/ericfranzon
> Twitter: http://twitter.com/SemanticWeb
>


-- 

Regards,

Kingsley Idehen	
Founder&  CEO
OpenLink Software
Company Web: http://www.openlinksw.com
Personal Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca handle: @kidehen
Google+ Profile: https://plus.google.com/112399767740508618350/about
LinkedIn Profile: http://www.linkedin.com/in/kidehen








Received on Friday, 18 May 2012 18:56:02 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 18 May 2012 18:56:02 GMT