I first found out about the Wolfram|Alpha search engine on Monday, courtesy of Wired. I loved it. I thought it was exciting, full of loads of potential and in no way a Google Killer, as so many new search engines are expected to be. As I grew up with a father with an insatiable desire to know everything there is to know about everything, I sympathized with Stephen Wolfram’s unrealistic but admirable ambition to delve into all the knowledge of the world.
And I liked the way article author, Steven Levy, described Wolfram|Alpha as a “powerful computational engine”. Although I had serious doubts about the part where he said that it “can effortlessly answer queries that no one has asked of a search engine before”, primarily because nothing in life is effortless (not even sleep) and I’m fairly certain that between all of the internet users in the world, many of whom are downright nutjobs, we’ve asked search engines some pretty interesting questions. Wolfram qualifies his search engine’s abilities by saying that it provides quantitative answers to questions. Quantitative – as in quantities – as in amounts – as in absolutely no semantics whatsoever.
And then I read Danny Sullivan’s review of Google Squared, and I also found that quite exciting, in a completely different way. To paraphrase Mr. Sullivan, Google Squared uses a spreadsheet-like format to provide answers to any search query. “For any search, Google examines the pages that it finds and tries to figure out what’s the best way to tabularize the information — IE, to put it in a structured, orderly form,” is how he describes it. And then he provides some screenshots, which are kind of cool.
Here’s a search for “baseball stadiums”:
For those of us who aren’t US citizens and who don’t get baseball at all, it looks pretty comprehensive, and Danny didn’t appear to have any issues with the results either. It also allows you to dig a bit deeper if you’re not entirely satisfied with the results by mousing over names and items in the columns, and it lets you add information that you think it missed and edits the results and all sorts of cool things.
It’s not perfect, but neither is Wolfram|Alpha. Both still have niggling issues to contend with before they are properly launched and both will have many more issues to contend with after launch when the real problems make themselves known via irate users. What do they have in common? Not semantics, that’s for sure.
So when I saw the title for Stan Schroeder’s article on Mashable “On semantic web: what it and what it never will be?” it never entered my head that it would be about the two very new, not even launched products mentioned above. You can imagine my surprise.
According to Stan, people use the term ‘semantic search engine’ to talk about two unrelated concepts: “adding metadata to the structure of the web, which should increase the overall organization of the bulk of data that comprises the Internet, and help machines (search engines, for example) find, share, and combine this data in a way that makes more sense to humans” and using conversational-speak when trying to get search engines to give relevant intelligible answers. He stands by the former and shuns the latter, which I agree with. I’ve often written on the inherent dangers of trying to devise machines capable of intuitively understanding human beings. And it gives me nightmares.
But here’s the thing, he says that Wolfram|Alpha pretty much epitomises the second concept of semantic search because people try to ask it intelligent questions. To which I respectfully say, Bollocks. People will ask it specific questions in the hopes of getting very specific answers, and if it works as it should, they will go away happy. If they don’t, well they can always rephrase or try Google Squared. He gives a couple of examples that he believes proves the futility of Wolfram’s search engine when compared to Google’s traditional search:
Which has a bigger gross domestic product, Spain or Canada?
What was New York City’s population in 1900?
When did the sun rise in Los Angeles on Nov. 15, 1973?
How far is the moon right now?
If I eat an apple and an orange, how much protein would I get?
Spain Canada GDP
NYC population 1900
Sunrise Los Angeles
Moon Earth distance
Apple orange nutrition protein
And? He’s not talking about semantics, he’s talking about wording and word order. And he touches on stop words, about which my esteemed colleague, Brett Pringle, just wrote a very interesting and very informative post. He talking about refining search queries, something I’ve also written about on several occasions, not semantics. Seriously. They’re not the same thing.
He concludes by saying that the semantic web will help us get better search results, and emphasises that trying to make search engines understand human speech is largely a waste of time. He says, “Computers and the web will not adjust to the way people talk. People will adjust to the way computers talk,” which great and I totally agree. But that’s not what Wolfram|Alpha and Google Squared are about, at least not my opinion.
I would love to hear what everyone else thinks. Am I a total moron, semantically speaking, or do I sorta kinda make some sense?