10 Tests for Google Squared

I’m playing around with Google Squared. This should be the type of thing that is just up my street – mining the web of data for multiple signals to form structured views of data.

Test 1: ‘small boats’ gives me:

• Kayaks – ok, reasonable
• Calendar – huh? location=Building ASP.Net Web Applications, Length=108 minutes
• Bolger boats on the web
• The boatbuilding webring
• West Wight Potter Web Page
• Canoes
• Accessories – huh?

Um – ok, how about something simpler?

Test 2: ‘sailing boats’ this gives something at least coherent, but it appears to have a hard time distinguishing between types (e.g. the class or make of boat) and instances (thus it provides ‘locations’ for boats which range from countries to specific zip codes).

Test 3: ‘netbooks’ again, something a little interesting, but it fails with its main value proposition giving processors as {AMD, 416 MHz, LGA775 Socket, Intel, …} close, but not really useful. Also the items in the table are brands (Dell, HP, Lenovo) not models.

Test 4: ‘scottish regions’ (regions are the main administrative areas in Scotland). Here the results are cities, so close, but no cigar.

Test 5: ‘british overseas territories’ – finally a reasonable results giving names, images, descriptions, capitals, currency and language.

Test 6: ‘plants native to the pacific northwest’ – very poor {gardening, botany, Hardcover, Your account, Add your first tag, Share your own customer images, paperback}.

Test 7: ‘novels of o’brian’ – excellent results.

Test 8: ‘jedi masters’ – oops {Obi-Wan kenobi, Star Wars: The Clone Wars, Mace Windu, Star Wars Episode VI:return of the jedi, Jedi Temple, Kit Fisto, Younglings}

Test 9: ‘movies of brad pitt’ – not bad, a reasonable list of movies with Director and Language fields. Note that it doesn’t attempt to normalize ‘USA’ and ‘United States’ suggesting that the system uses very superficial representations internally.

Test 10: ‘progressive rock bands’ – fail {Souther rock, Arena rock, Boogaloo, …'}

Overall, I find the idea intriguing, however the results currently seem very preliminary. I’m guessing that this was surfaced to the public for some political or strategic reasons rather than as an example of Google magic. Here’s ReadWriteWeb’s take,

Comments

[FD]

So, the idea was "let's release an obviously beta product hoping to steal the publicity thunder from the competition" (that, in the case of WolfranAlpha and Bing, actually showed value and innovation in their products.

Does this Google strategy make sense to you?

IMHO, I think it decrease the value of the brand Google.

[Tony Hirst]

It is possible to crate your own squared search grid using data sources either of your choosing - or arbitrary Google results - in a Google spreadsheet. eg "Using Google Spreadsheets and Viz API Queries to Roll Your Own Data Rich Version of Google Squared on Steroids (Almost…)"
http://ouseful.wordpress.com/2009/06/05/using-google-spreadsheets-and-viz-api-queries-to-roll-your-own-data-rich-version-of-google-squared-on-steroids-almost/