The Google search engine is powerful, but not all-knowing. Every month, Google handles 100 billion queries, returning the results in a matter of microseconds. But as it turned out, quite a large percentage of these requests requires more sophisticated processing, as occurs for the first time. Daily, 15% (500 million) of all queries are unique and have never met Google. This high percentage is stored for the past 15 years of existence, according to John Wylie, a leading developer of Google Search.
In order to reduce this figure, constituting 15 percent of the new, not previously encountered queries, Google adds to the database 20 billion ads per day, in search of new information.
A key step in reducing the percentage of queries not previously encountered and to provide answers, not just lists of links, is the creation of knowledge networks (Knowledge Graph). This extensive database, similar to the human brain, contains information about people, events and objects, as well as the interactions between them. At the moment, in the Knowledge Network contains 570 records and 18 billion relationships between them.
"Still a lot of what this network does not know. We collected only 1 percent of the information," - said John Dzhennandrea, who heads the development of the Knowledge Network.