Google Knowledge Graph

Ever since online search began in the 1990’s, Google’s Knowledge Graph is the latest attempt at bringing in a lot more information, without the user leaving the search page or clicking on a link.

Launched last week in the US, Google Knowledge attempts to provide direct answers to queries like ‘Taj Mahal’ or ‘T20 Cricket’. It’s a big attempt to move away from giving just information to a convincing answer.

The difference is that Knowledge Graph, on the right panel, adds boxes and information full of Wikipedia-like information to the search results.So lets say in the past, a query like Taj Mahal was understood as just two separate words. But Taj Mahal could have much more in depth meaning – one of the Seven Wonders of the World, or it could mean a casino in Atlantic City, New Jersey or it can be even an Indian restaurant in the US.

So What’s different?

If most people earlier looked up while searching Taj Mahal, the Knowledge Graph will provide a brief outline of the monument, its location, who built it, when and why and other monuments the searcher might be interested in, say Qutab Minar. While the casino and the restaurant will be provided as links. 

What is interesting is that the graph draws on public sources including Freebase, the CIA World Factbook and Wikipedia to provide answers. Besides, the search algorithm decides what information to show based on collective wisdom of other Google users. 

It Google two years, and the final push came in when Google  acquired Freebase, in July 2010, a massive open source database of information about almost anything including books, movies and music. 

So a query ‘Sachin” could mean the cricketer, the yesteryear Bollywood actor or your friend on Facebook. However, the Knowledge Graph, based on history of queries( from other users) will opt for Sachin Tendulkar and provide facts about him including his date and place of birth, the 100 centuries, details about his family and so on without users having to leave the page. 

So why this change?


The big picture would be to create stickiness (people stay longer on the search page), provide more focused advertising and get more people to click on ads. With a $60 billion or so of annual revenue expected to come in 2014 from ads that appear next to search results, this does provide the right path.

Microsoft made a similar attempt, to go beyond the 10 blue links, when it acquired Powerset (a semantic search engine) in 2008. Since then, Microsoft’s Bing has been offering more relevant search results and nibbling at Google’s market share for search. And rival Yahoo claims that it started talking about Knowledge Graph type results about eight months back. 

While Google has got a step closer to giving answers and not just information or links, it has limitations. According to IDC, the search haystack today exceeds 1 zettabyte of information – one followed by 21 zeroes. 

In simple terms, if all the six-billion-plus people in the world posted messages on twitter continuously for 100 years, that’s the information pile, they would create. But Knowledge Graph relies on a very limited data set and is good only for more definitive search of say a celebrity, a movie, a monument, a sports team or a historical event. 

The graph is available in the US only as of now and reviewing it Rafe Needleman, a technology writer, says: “Search for concept like ‘mortgage’ and you get nothing from the Knowledge Graph. Look for news item ‘Facebook IPO’ and there’s nothing in the Knowledge Graph. 

Though, search for a mainstream movie or a celebrity gossip and you will find loads of information cross linking directors, actors and similar shows or films.”Definitely a long way ahead.