Another day, another way to search the web
Jun. 8th, 2009 01:58 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
jducoeurRemember me mentioning, a couple of weeks ago, how Google would likely come out with a competitor to Wolfram Alpha? It's here, albeit only in very early-draft form.
This week's interesting gadget is Google Squared. The idea is essentially that, instead of just presenting search results as links, you can have Google try to decompose the data that it finds into columns, for quick and easy comparison. Ars Technica has a dreadfully funny article about how bizarre the results can be -- unlike Wolfram Alpha, which very carefully tends each little bit of data to make sure that it works right, Google is attempting to guess what is relevant and interesting for this query, and sometimes fails spectacularly.
Still, used properly, the thing looks like it has potential. For instance, the most obvious first query I thought of for it was "restaurants in Burlington, MA", and I got roughly the results I was expecting: one row per restaurant, with a picture (sometimes right, sometimes not) of it, its address, and its cuisine. It also included a "Specialty" column which seemed to be mostly junk, but the nice thing about Squared is that it lets you refine the query yourself -- so I could kill the irrelevant "Specialty" column, and add "Hours" easily. Knowing a particular restaurant I wanted to compare (I used Landana), I could just type that as a row, and Google looked up all the column values. And Squares can be saved and shared once you have the query refined to the point you like.
It's not a world-shaker: it needs a *lot* of improvement before it's even going to be worth using for real. But I do think it shows a lot of potential as another blade in the Web's utility knife, especially as the Semantic Web grows, and gradually adds a bit more structure to the Web's uneven melange. Particularly for cases where you have a query that you expect "square" results for -- lots of rows, each of which has distinct and fairly consistent data fields -- it's likely to eventually become a tool of choice...
This week's interesting gadget is Google Squared. The idea is essentially that, instead of just presenting search results as links, you can have Google try to decompose the data that it finds into columns, for quick and easy comparison. Ars Technica has a dreadfully funny article about how bizarre the results can be -- unlike Wolfram Alpha, which very carefully tends each little bit of data to make sure that it works right, Google is attempting to guess what is relevant and interesting for this query, and sometimes fails spectacularly.
Still, used properly, the thing looks like it has potential. For instance, the most obvious first query I thought of for it was "restaurants in Burlington, MA", and I got roughly the results I was expecting: one row per restaurant, with a picture (sometimes right, sometimes not) of it, its address, and its cuisine. It also included a "Specialty" column which seemed to be mostly junk, but the nice thing about Squared is that it lets you refine the query yourself -- so I could kill the irrelevant "Specialty" column, and add "Hours" easily. Knowing a particular restaurant I wanted to compare (I used Landana), I could just type that as a row, and Google looked up all the column values. And Squares can be saved and shared once you have the query refined to the point you like.
It's not a world-shaker: it needs a *lot* of improvement before it's even going to be worth using for real. But I do think it shows a lot of potential as another blade in the Web's utility knife, especially as the Semantic Web grows, and gradually adds a bit more structure to the Web's uneven melange. Particularly for cases where you have a query that you expect "square" results for -- lots of rows, each of which has distinct and fairly consistent data fields -- it's likely to eventually become a tool of choice...