ResearchBuzz

Google Books announced today that it’s added a “Common Terms and Phrases” cloud on the overview page for some of its books. We can’t call it a tag cloud, because these aren’t tags, so I’ll call it a term cloud.

For an example let’s look at Songs of the Cattle Trail and Cow Camp By John Avery Lomax . A little past halfway down the page you’ll get a cloud of commonly reoccurring words that looks like this:

Term Clouds on Google Books

Click on any of the words in the cloud and Google will take you to search results for that word in the full text of the book. In the case of this book the cloud was a bit misleading; “Whiskey Bill,” for example, appears several times in a song which is why it got such a big tag.

Taken by itself, I think this is interesting. A fun toy. What made me really raise my eyebrows and go “whee!” when I read it, however, was the fact that the New York Times last week announced that it was going to release its thesaurus. That’s over a million terms.

Wouldn’t it be neat if you could go to the New York Times and do a keyword search, then get the thesaurus terms for the your search results and apply them to a Google Book search? Or how about doing that in reverse? I’ve found a book in Google Books I really like, and I want to learn more. I can look for more books on Google or I can search for terms in Google’s book cloud against the New York Times thesaurus and get current articles on my topic of interest.

I mean, I could do that if anybody ends up making a tool to do that.

After poking around a little I could not find a way to search Google Books just for items in the term clouds. Aw, c’mon Google…

Even after all the hype, Wolfram|Alpha (http://www.wolframalpha.com/) didn’t do all that much for me. Yeah, I liked it, thought it was neat, and had used it a little bit as a reference engine, but it just hadn’t turned me into a rampaging evangelist. Then I found I was using it a lot because of the functionality that goes beyond just giving answers. It’s kind of edging its way into my daily search chores.

Like I said, I started using it as a reference engine when I had basic questions. And the nice thing is when you need to get the same data point for several different items, you can list all the items in your query — separated by commas — and WA will give you a table of the results.

(If you’re having a hard time visualizing what I mean, search WA for population of durham county, population of guilford county, population of carteret county, population of harnett county.)

This seems like kind of a silly example but wow was it convenient when I quickly needed sets of data for certain counties to compare population densities.

Then later on I was working on a finicky spreadsheet that uses lots of calculations and a huge lookup table. I could not run calculations through the lookup table and know at a glance that everything was being calculated as it should. So I just opened my browser, called up WA, and had it do algebra for me when something didn’t look right. If you query WA with, say, 712x = 3.15, WA will solve for x. It told me when percentages were being rounded, when maybe I needed to add more detail to my lookup table, etc.

I had concerns about WA not being personally useful and totally missed that it would end up being so handy. I also discovered a fun WA game: enter in several different items that could be part of a set, just not the same set. So you might want to combine planets and colors: saturn green blue mars yellow earth. WA will go a little nuts trying to put all the items into one set. Then it’ll give up and give you additional options for those query words which might be relevant, which is how I discovered WA had data on the Yellow River and that WA considers “Mars yellow” a color.

It’s amazing to see all the search engine development that’s been taking place in the last couple of months. It’s like 1999 all over again, in a good way. Google Labs is humming with all kinds of neat stuff. When Google Squared came out earlier this month, I was pretty excited, but I think right now I’m at “It’s … okay.” A little disappointing.

Google Squared is available at http://www.google.com/squared. Here’s basically how it works. You search for something that will return a lot of things, like planets or cartoons. Google then returns a search result page that looks like a square (surprise!) and which contains items that hopefully fit what you were searching for. Each item has several columns of data, and you have the option to add more.

So I did a search for candy bars. I got a page of results and most of them were candy — not candy bars, but okay. Except for one of them which was for “Ion”, and while the description in the square contained information about a chocolate bar, the picture for the column was of a battery. (And I’m pretty sure it wasn’t a chocolate battery.)

Annoyed, I did another search, for monty python, and realized two seconds later that was WAY too general. I did another search for Monty Python cast and this time I actually got a pretty good square, with a list of people affiliated with Monty Python. The first page of information was great, while the second one was starting to go a bit far afield. After a couple of minutes I realized that Eric Idle was missing. Eric Idle! How could you list Neil Innes and not list Eric Idle?!

Feeling topical (and wanting to throw Google Squared a softball) I searched for the Jackson Five. Google Squared actually returned me a front page with six Jacksons (Randy Jackson) and Diana Ross. The first result for this square was Michael Jackson, and it would be easy for someone to view the picture that’s been included as the representation of his person and find it offensive.

Anyway, let’s go back and do another search. If you do a search for candy you’ll get an idea of the kind of thing Google Squared can do. Here’s a screen shot:

Looking at a search result.

As you look at it you’ll see that each result is in a row, and each row has several columns of data. My candy searches seemed to default to price data. You can remove any column and add more columns if you like. As you can see in the next screenshot Google will make suggestions for the columns. As you also see some of these suggestions don’t make any sense.

Google Squared Suggestions

Google does say that if you don’t like the squares that Google is offering you can start with an empty square. I was still thinking about candy bars, so I started with an empty square and entered Baby Ruth and Almond Joy and Chick O Stick. My idea was that I could enter the kinds of things I was looking for, and then Google Squared could complete the list for me with similar items.

What was I REALLY thinking of? That’s right, Google Sets, which may be my favorite Google Labs product that never graduated. That’s what I wanted. And Google Squared doesn’t do that. It’ll make suggestions once you’ve started your square, but it won’t allow you to enter a few suggestions and then automatically build a square for you…

.. at least, not in a way that’s easy to find. Google Squared actually has the Google Sets functionality built in, but the way I found it was to mess up. If you enter a query that Google can’t easily square (I liked wombat inarticulate milkcan), Google will give you an error and prompt you to enter a few items upon which it can build a square:

That way I was able to enter a few candy bar names and get an actual square that I liked.

One thing I noticed is that I was never able to get full nutritional information for any of these candy bars. It was always ordering information.

So on a whim I went to Wolfram Alpha and entered in a bunch of candy bar names, separated by commas. WA gave me nutritional information for all the candy bars on one screen, even asking me to clarify which version of Twix I meant!

Not placed in a square, and not as slick as Google, but oh-so-handy.

Google Squared is a neat idea. But the squares it comes up with are odd. Sometimes they miss critical bits, and sometimes they just don’t make sense. It would be intensely useful to have some kind of way — a keyword search? a switch? — that allows you to specify what kind of columns you want to see, instead of wading through suggestions. A bit more functionality, and it could be super cool.