Google Gets Semantic, Launches Knowledge Graph Starting Today
This “Knowledge Graph” is a two-year-old project that evolved in part out of Google’s acquisition of Metaweb in 2010. Google now says it understands 500 million entities and 3.5 billion attributes and connections.
When users search for a term that triggers the Knowledge Graph, they’ll see a box of information on the right-hand side of the search results page.
The boxes contain all sorts of information that’s specifically relevant to the search term. For instance, a results box for Leonardo da Vinci would have a brief description of him, his birth and death dates and his parents’ names, pictures of five of his most famous works and links to other artists that people often search for when they’re looking up da Vinci. It’s a lot like a dense and visual Wikipedia page.
When relevant, Google will ask users to specify what sort of entity they are looking for. So if you search for “kings,” the box might include disambiguation links for the Los Angeles Kings of the NHL, the Sacramento Kings of the NBA and the NBC drama “Kings.”
With all that information right there on the Google results page, users might be less likely to click through to other Web pages. I asked Google search engineer Ben Gomes about that, and he deflected the question.
“Our goal is help you explore a topic more,” Gomes said. “We’re providing you with a skeleton which we’re using to organize information. But if you actually want to find deep information around a topic, we have the Web pages to provide you with that information.”
At least on an interface level, Microsoft is on a similar track with Bing — where it just launched custom panels for results in 150 categories. But what Google is doing goes quite a bit deeper.
Gomes described the Knowledge Graph project as part of Google’s overarching “progression from data to information to knowledge.” He said that Knowledge Graph results will turn on for a “significant fraction” of Google queries — about the same as local results.