Searching for the 5 W's

Written by Adrian Holovaty on October 27, 2002

Googlism uses Google to give you one-sentence answers on "who," "what," "where" or "when" a particular term is.

Most of it is silly. For example, a "who" search for "Paul McCartney" results in a wacky list of descriptions ranging from "paul mccartney is a great musician" to "paul mccartney is a futile gesture." Whatever that means.

But some of it is dead-on. For example, a "who" search for my name returns this single result:

adrian holovaty is assistant database editor and product developer at the atlanta journal

Not bad at all -- especially for something created just to be a "fun tool."

I mention this site (which I heard about via Simon Willison) because I think this technology, when improved, would be an intuitive way to get news. It'd be great to be able to do, say, a "who" search for "Alan Greenspan" or a "what" search for "World Series," and get a results page with meaningful content instead of a laundry list of article links.

For the "Alan Greenspan" search, perhaps the result page would give a short biography of Mr. Greenspan, plus a short summary of how he has been in the news lately. (With links to more detailed stories, of course.)

For the "World Series" search, perhaps you'd get a synopsis of the baseball series so far, with historical information and, again, links to more in-depth content.

The questions that Googlism lets you pose, and the questions that news stories address, are a natural fit. Journalists in the audience will note "who," "what," "where" and "when" are four of the somewhat cliché five W's and one H -- the questions that effective news stories should answer. (The remaining two questions are "why" and "how," but those are beyond the scope of a simple text-search engine, for now.)

Comments

Posted by Ben on October 27, 2002, at 12:35 p.m.:

I love the googlism thatis returned for my name ben meadowcroft.

"ben meadowcroft is correct"

Enough of that. The googlism reminds me of the Google sets lab site (see labs.google.com to get the link for it. Basically how it works is you enter a set of data (think mathematical sets) and then attempts to predict a larger superset that is related to the original data. I think that this too could have implications for news sites, this kind of algorithm could be used to determine related stories automatically (and remove the need for people to search through the archives for related stories).

When I think of news sites I am reminded of a comment an IBM employee made about IBM wanting o employ T-shaped people, menaing people who were deep in their speciality, yet also having a broad base of knowledge. News sites too could adopt a similar strategy, When a person browses to a story offer them a deeper insight into the elements story (RE: Alan Greenspan) as well as broader coverage of related articles (which would also have their own depth, but this depth would only be apparent when the story was browsed to not beore)

Any way enough of my unstructured thoughts, suffice to say that I think the sets idea is a good one. After all traditional search falls down when the words a person is looking for are not actually in the article, with a set search, if the word is not contained in the article then a superset of words is searched for, hopefully bringing returning articles related to the orignal search.

Posted by Kevin Laurence on October 28, 2002, at 12:29 p.m.:

Don't you think that even the sensible answers are strange responses to the question "Who"?

I mean when someone asks me who I am, I don't recite my job description! I tell them my name. I only describe my job when they ask me what I do for a living.

I know some will consider it pedantic, but I'm not sure I want to be defined solely in terms of my work.

Nor do I want to be defined by someone else's profession:

"kevin laurence is a principal in the intellectual property practice group of stoel rives".

Heaven forbid that anyone should think I'm a lawyer!

Posted by Dave on October 28, 2002, at 8:04 p.m.:

Googlism is missing in action. Note on site says "see you real soon!" Sorry I missed it.

Posted by Adrian on October 29, 2002, at 11:16 p.m.:

It's back.

Comments have been turned off for this page.