Google launches its Knowledge Graph, a step toward semantic search

Google unveiled Wednesday on its Knowledge Graph (literally graph of knowledge), a new interface element that aims to answer questions directly from the user from the results pages of its engine.

Organize the world’s information? The famous creed Google has perhaps never been clearly assumed that with the launch of “Knowledge Graph”, a new tool that capitalizes on engine indexing capabilities to suggest, automatically, the answers and relevant content to the user based on his research. With a size of nuance: the content, links, images or maps are not simply offered as links, but displayed directly on the results pages of Google, in the right column.

An example? Imagine you launch a search on Vincent van Gogh. There is a good chance that you are primarily interested in two types of data: the biographical information related to the life of the painter, or his most famous works. With this “knowledge graph”, which builds on all previous searches, Google is able to determine your most likely interests, and return, alongside traditional search results, the more elements capable of meet your expectations.

Why speak of “graph”? The search engine explains that this new feature is based on a database created by it, which lists more than 500 million objects of any kind (people, places, works, etc). The real value proposition, however, not simply lies in these objects, but in the relationships that are woven between each of them, in order to assess more precisely the aim of the research. These relationships are believed to number 3.5 billion today.

“The Knowledge Graph helps us to understand the relationships between things. Marie Curie is a person in the Knowledge Graph, she had two children, one of whom also received the Nobel Prize, and a husband, Pierre Curie, who ran for a third Nobel Prize for the family. All these elements are linked in our graph. It is not just a catalog of objects, he can also shaping these internal relations. It is the intelligence between these different elements which plays a key role, “illustrates Amit Singhal, vice president in charge of engineering at Google.

According to Google, the added value of the Knowledge Graph is established at three levels. The first is of course to help the user find what he really seeks (Taj Mahal it is the Indian palace or the restaurant around the corner?). The second is to directly display the desired information (birth and death dates for a person known example), without the user having to perform an additional click.

The third, perhaps most ambitious, is based on the presentation made by Google, to promote the discovery and further research by automatically suggesting additional answers of those who respond to the basic request of An internet, combining semantic analysis to the principle – sovereign Web – serendipity.

“We always thought that the perfect search engine would understand exactly what you mean, and return you exactly what you want. And now we are sometimes able to answer your next question before you asked that, because the facts we show are enriched by elements sought previously by other people, “says Amit Singhal.

The new motorized interface elements by the Knowledge Graph is currently being deployed on the English version of Google. They are arriving later on international versions of the web portal, as well as its various mobile versions, where interest would be even higher depending on the engine.

Besides the promise of improved relevance and effectiveness, Google should earn a dimension “media” extra, keeping longer users on its pages instead of only serve as a gateway to other sites. In its way, it also responds to Facebook which, with its Social Graph aims to map all social interactions: the Knowledge Graph neither seeks nothing less than knowledge.

The video below summarizes and illustrates (in English) the scope of the project.

Edited on 11/07/2012 at 4:27 p.m.

Comment (28)