Finally, Google’s search algorithm is getting a huge boost from a source we can actually trust.
No, it’s not your dimwit friends and their Facebook shares. It’s Knowledge Base, a semantically structured body of data 500 million items strong and growing.
With Knowledge Base plugged into Google Search, Google is going to get a heck of a lot better at knowing what things (not just what words) you’re searching for and delivering better results — more useful, more visual results.
How Google’s new brain works
Briefly, Knowledge Base is Google’s attempt to change your search experience from strings (of characters) to objects, whole entities with collections of their own data.
For example, yesterday, Google could parse your search for “roots” as five letters that might have a connection to a wide range of URLs. Today, Google knows you are probably searching for The Roots (the music group) or or Roots (the TV miniseries) or roots (the biological feature of plants), and it will help you get to the best results for that object faster.
Here’s a quick tour of how Google Search will look with Knowledge Base powering it:
Knowledge Base takes structured data from Wikipedia, Freebase, and other sources, including MusicBrainz (an open encyclopedia of music facts), the CIA’s World Factbook, and many other repositories of publicly available open data.
“Our overall mission is to organize the world’s information,” said Google Search project management director Jack Menzel in a phone conversation with VentureBeat yesterday.
“We’ve done some really clever things when it comes to understanding your query, but computers don’t really understand what people are talking about,” he continued. “To a computer, it’s just a string of letters.”
While Google’s parsing of misspelled and incomplete search terms as well as synonyms and related terms has been helpful to users, Menzel has been working to give Google’s search engine a more useful, more advanced understanding of search terms as interconnected objects in a database.
“As a search engine, all the results might look like good web pages, they all contain that word … but it’s not enough,” he said.
Why we’re in love with structured data
Knowledge Base has been a Google project at least since the company acquired Metaweb in 2010. The San Francisco startup was the maker of Freebase, a massive open-structured database. While Freebase was a starting point for Google’s foray into structured data, it was just the beginning.
Web companies are having a very public love affair with structured data these days. Basically, companies like Google, Microsoft (with Bing), and Facebook (with Actions) are attempting to understand not just the letters and words you type into their systems, but also what those letters and words mean and how they’re connected to URLs, to each other, and to sets of facts.
“We look at all the publicly available information on the web, and we try to reconcile them into a representation of the real world,” said Menzel. “It’s a huge challenge, and we’re just getting started.”
But, the Googler noted, even at its initial launch to the public, Google’s structured database contains 500 million people, places, and things; and these objects are connected by 3.5 billion attributes. Menzel thinks this gives Google’s web search a competitive advantage over Bing.
And because of how those attributes are structured, Google search is now able to return fascinating chunks of information for many kinds of queries — think of it as rich snippets on mega-steroids.
“When you do a query for a specific, real-world thing, we can synthesize a summary on the fly based on what the most relevant information is,” said Menzel. “That acts as a map of all the information that’s available to you on the web.”
For example, if you search for “roots,” Google will ask you if you mean “The Roots” the musical act. If, in fact, you do mean to search for that group, Google will quickly pull up a slew of related data — stats, members, shows, pics — all neatly organized in a scannable table for you. Think of it as a mini-infographic each time you search for something.
Here’s a video Google made to show off some of the new Knowledge Base features:
The “Wikipedia Hole” is coming to Google
And the searches just keep coming, too. “When you’re doing a research task, you very rarely want to do just one query,” said Menzel. “We help people create a map of the topic they’re interested in so they can more quickly explore the information.”
Menzel and the Google web search team call this “serendipity.” You or I might call it a “rabbit hole” — a dark, delicious tunnel of information, a bottomless pit of trivia from which you may never emerge. This happens a lot to the casual Wikipedia reader, and while it can suck up a ton of time, it can also yield some fun facts and new knowledge for your sponge-like brain.
Also, just like Wikipedia, Google has built in some tools for reporting incorrect information. “This is a tremendously ambitious task that’s really, really hard, and the world is constantly changing … so we’ve built a feedback mechanism so that in two clicks, you can report any problems you find,” Menzel told us.
The long and winding rollout
The new search features are launching today and will continue to roll out over the next few days for Google search on all platforms and devices — in English only, for now.
“Ooh boy, languages are hard. And understanding how to rank this content internationally is hard,” said Menzel. Other languages and regions will be coming soon … ish.
“As you can understand more about what people and web pages are talking about, you can imagine doing queries that are pure science fiction,” Menzel continued, “like finding a Lady Gaga concert this year outdoors in warm weather. Today, that’s crazy. That’s going to take a long time to figure out. With Knowledge Base, we’ll be able to answer that very complicated and nuanced question quickly.”
Finally, we asked Menzel how Knowledge Base — that big, slick, data-driven machine — was going to mesh with Search Plus Your World, Google’s social approach to search, which launched earlier this year.
Ultimately, he told us, Google web search is just about getting you quickly to the best possible information.
“Sometimes, that best information is the private information that only you have access to, like the Search Plus Your World information,” he said. “And sometimes it’s Wikipedia, and sometimes it’s sports data, and sometimes it’s the weather. I don’t want to belittle the importance of social data, but knowledge overall does have a larger scope.”
Image courtesy of alpahspirit, Shutterstock
VB's research team is studying web-personalization... Chime in here, and we’ll share the results.