Source: Wired, Jan 2013
The larger goal of the Knowledge Graph is to enable computers to understand the world the way humans do. … The Metaweb and now the Knowledge Graph, has been absorbing the world’s structured databases.
The work of the semantic graph is to make the connections that traditional search might overlook. “You’d be surprised at how many times there’s a serendipitous link between two different things,” he says. “It can be hard to describe for computers to understand that: what’s the relationship between Einstein and Gandhi? They were both pacifists in later life.” This might be a common search inquiry, he suggests, but the computer cannot work out why.
As much as trying to know everything, the Knowledge Graph is about trying to work out what you want to know, parsing the disambiguation (“did you mean?”) and filtering noise.
With the Knowledge Graph, Google has taken a different step towards the future of search: providing answers, not links.
The Knowledge Graph also challenges the organisational hegemony of Google’s dozen blue links. “The web is very top down, in terms of links and anchors,” says Giannandrea. “What you can’t really do in a web browser is look at a page about a particular play and think, ‘What other plays should I consider?’
We need to be able to go sideways through human knowledge.” This is figurative, but also literal: entering ”London bridges” on Google, one now sees an image carousel of London’s most significant bridges, arrayed horizontally. This is possible because those bridges have been encoded as entities in the Knowledge Graph. But what happens when that knowledge is not encapsulated in structured databases, when it’s not a piece of text, or even when the subject of one’s search is something the user is looking at in the moment?
How closely these words lie near others will help determine context and relevance.
Dyson notes that Bill Gates told her: “the future of search is verbs.” People, the argument goes, want search to do things, not just suggest things. With the Knowledge Graph, Google is building a world-historical collection of nouns.
When it’s suggested to Singhal that the future of search may not really be “search” at all, but some as yet undefined process, his answer is quick: “I won’t get hung up on words,” he says. “You can call it whatever you want. This is what the human brain would like to have by its side, when you’re seeking information, or sometimes information comes to you without your seeking it.“