Google Licenses Australian Algorithm

Australia rules — or at least it does when it comes to Google’s latest move. The search engine gained a new employee and the right to use the algorithm he developed at the University of New South Wales. What does it do? And what might Google be getting out of the deal?

For something that may have technically happened six weeks ago, it certainly set the press buzzing. Then again, how often does a 26-year-old graduate student working on his PhD in Australia get hired by the world’s best-known search engine? And that for his work on an algorithm that many insist will revolutionize search?

Anyone who has been covering search engine news for a while will tell you that claims of revolution are usually (though not always) overblown. See the word “revolution” paired with search technology and you’d be smart to mentally substitute “incremental improvement.” But even that much holds the promise of making search faster, easier, and more accurate – and that’s a big help when you spend a lot of time searching.

So who is this graduate student, and why is Google so interested? His name is Ori Allon, and he hails from Israel. When Google found him, however, he was working on his doctorate at the University of New South Wales. A press release from the university dated September 2005 describes the technology, dubbed the Orion search engine, in far more detail that Allon is legally permitted to go into now. Though employed by Google, he can’t even reveal his job title.

So what could a graduate student working for a year in a fourth-floor room with five servers accomplish? Allon came to Sydney to work on a project run by Dr. Eric Martin, of the university’s computer science department. Martin’s project aimed to build a better search engine. And according to the scientist, “What others would have taken two or three years to do, [Allon] did in six months.”

The project is supposed to be finished sometime in the next 12 to 18 months. The university will retain ownership of Orion, because it was developed within its PhD program. It will probably receive royalties from Google for licensing the technology; some analysts figure it could run into the millions. Nobody really knows how much because nobody who has information about the actual deal is talking.

{mospagebreak title=What Does it Do?}

It’s all about letting the user find out for himself what is most relevant. The Orion search engine hunts down pages where the content covers a topic that is strongly related to the searched-for keyword or phrase. It returns a section of that page. If it did a really good job, a searcher might find the answer to a question right in that section of text, without having to visit the site itself.

That’s only part of what Orion does, though. It also lists other topics related to the keyword, letting the searcher choose ones that are most relevant. If any particular broad topic is like a tree, related topics are branches off that tree, and Orion can help a searcher find the twig they were looking for – or even get a better picture of the whole forest. The university’s press release about the technology describes it as “offering an expert search without having an expert’s knowledge.”

It then provides an example that many of our readers in the United States should be able to relate to from their elementary school days. “Take a search such as the American Revolution as an example of how the system works. Orion would bring up results with extracts containing this phrase. But it would also give results for American History, George Washington, American Revolutionary War, Declaration of Independence, Boston Tea Party and more. You obtain much more valuable information from every search.”

You can easily see how this kind of search engine would save a lot of time. How many times have you found yourself clicking through to a website to find your answer, not finding it, then clicking back to your search results to find a site more likely to contain your answer (and repeating this process several times)? Rather than going through that process, Orion could give you all the answers you need to a query on one page. And if your answer isn’t right on that page, you still have the option to click through to the site itself. The expanded extracts from the websites should give you a good idea of where to go to get that answer.

It’s not perfect, of course. As already mentioned, it’s not quite ready for prime time yet. And by its very nature, Orion is likely to work better for people looking for information rather than something to buy. Then again, content is supposed to be king online, right? Even if it doesn’t revolutionize web searching, it just might revolutionize homework, as students find the answers to their questions online a lot faster and easier.

{mospagebreak title=More About the Project}

Licensing the technology to Google, after also being reportedly courted by Yahoo! and MSN, is not a bad return on investment. The project itself was funded by a $150,000 grant from the Australian Research Council. And I have to give them credit; you really need to know how to read the academic language to find the promise in this project.

It was titled “RichProlog, a System for Deducing, Inducing and Learning in the Declarative Programming Paradigm.” And I know I could never work for the ARC now. The project aimed to “contribute to bridge the gap between learning and logic, theoretically and practically,” “extend considerably the scope of the declarative programming paradigm” and “build a system that can be used to solve learning or discovery problems as encountered in Artificial Intelligence.” Only the third statement suggested something related to search engines to me, which probably means I should stick to reporting this stuff rather than trying to write a search engine (or a letter requesting a grant, for that matter).

The return on that $150,000 investment could mean a steady stream of royalty payments the University of New South Wales if Google actually incorporates the technology into its search engine. Some analysts think the Orion technology could indeed become the basis for the next generation of search engines. The idea is that it will improve not only the relevance of the results, but give searchers a better idea as to whether those particular results have the answers they need, thanks to the more detailed sample of each file.

This has certain implications for web publishers, one of which I already mentioned – the likelihood that searchers will have less need to visit the website that has the information they need. When asked about this point in an online interview, Ori Allon gave a thoughtful response. “This is a very good question. I don’t envision that Orion will completely eliminate the need for going to actual web pages but rather expedite the search process. It could also result in the user going to more web pages since they will be alerted to other keywords related to their search.”

{mospagebreak title=What’s in it for Google?}

Allon has been quoted as saying that Orion would improve “the speed and focus of internet searches with is, as we all know, an invaluable service.” That much is obvious, and probably reason enough for Google to latch on. But there are other points to consider here.

Think for a moment just how wonderful a search like this would be when going through books. It’s practically designed for anyone looking for information rather than, say, the website of a particular company. When you add in the fact that it turns up related keywords and topics, you have the makings of a good research paper assistant. Given that Google is working on a project that involves digitizing the content of many volumes of books, you can see how this kind of algorithm would be immensely helpful.

That helps the academic community, but there are also good, commercial reasons for Google to want this technology. If it has the side effect of making people stay longer on Google’s site, rather than clicking off to find their answers, then Google can serve more ads. Remember that well over ninety percent of Google’s revenue comes from advertising.

While MSN and Yahoo! have portals with their own content to help make their sites “sticky,” Google doesn’t have a portal unless you’ve customized the home page (though it certainly has a variety of services, from groups to blogs to email, and more coming out every day). As one analyst wondered, “Could Google be looking to create a mega-site that could provide the answers – and the information – to absolutely anything?”

You probably think that sounds a little grandiose. But remember the search engine’s mission statement: “Google’s mission is to organize the world’s information and make it universally accessible and useful.” Licensing the Orion technology could bring Google one step closer.

[gp-comments width="770" linklove="off" ]