Google’s First Step To Becoming A ‘Knowledge Engine’

The algorithm used by Google to search the internet will be the subject of a major upgrade today in an effort to contextualise search results.

Click to enlarge

Currently Google’s search recognises the make-up of letters and words, and then scours the internet in search of resembling content, but the search engine fails to understand what it is you’re actually searching.

Google wants to change this by imbuing its search engine with the intuition needed to understand what is being searched for. Take The Australian’s example of the “Taj Mahal” for instance: Google will be able to identify it as the white marble mausoleum in Agra, a Grammy Award-winning musician or a US casino.

Once Google discerns the difference between the three, it can then use its new Knowledge Graph to display parallel results, links and summarised information.

For example, a search of director Quentin Tarantino will generate a biography, his movies and relevant people who have also been searched for, which in this case includes director Robert Rodrigugez, Martin Scorsese, and Tarantino’s muse Uma Thurman.

The knowledge graph used to drive the new feature is a data base that consists of 500 million people, places and other entities, and recognises more than 3.5 billion relationships between them. Google acquired the database in 2010 when it purchased Metaweb technologies, who were working on ‘Freebase,’ but at the time that housed just 12 million searchable entities.

Google’s senior vice president of engineering, Amit Singhal, claimed Google also ascertained information from the CIA world factbook and Wikipedia.

Also announced yesterday was Search’s ability to generate relevant Gmail results. A user who searches for, say, an airline could have their previous bookings generated; however, this feature is to be trialled with one million users first.

Google revealed 3 billion searches happen each day and more than 230,000 webpages are added or revised each second on a typical day.