Mountain View, California–The internet’s most popular search engine just rolled out Google Hummingbird, an algorithmic update at the core of processing queries. Rather than parse search phrases by individual keywords, the new technology attempts to handle more complex searches like full questions.
Revealed by Amit Singhal, Google’s senior VP, in the garage rented by the co-founders of the search giant, Larry Page and Sergey Brin, Singhal announced that it was the product of fifteen years worth of work. “Our algorithm had to go through some fundamental rethinking of how we are going to keep our results relevant,” Singhal said.
Google Hummingbird Takes Flight
Though the engineer did state to reporters at the press conference the new technology would be able to process more complex searches, he did not reveal any of its inner workings. Singhal only elaborated that the new search mechanism would better understand concepts, relationships between different concepts, and more complex search queries.
Though Google has stated many times it employs 200+ algorithms in its search and ranking technology, actually little is known about specific parameters and mathematical processing. Most of what is known is the result of independent analysis, but hard data is closely guarded by the search engine.
The Next Step in Search
Google Hummingbird embraces what many call the semantic web, or what others refer to as Web 3.0. Sine the inception of search, keywords have been used to match queries with results. While that isn’t changing, there is now a genuine effort to more accurately return results based on concepts rather than just parsing words. For instance, when a person speaks about the word “fly”, it’s in relation to a concept–an insect, a mode of travel, or slang for part of a gentleman’s apparel. Search engines are attempting to “understand” queries much the way people process information.
Google is currently the most used search engine on the internet, processing 67 percent of all queries, which account for more than 100 million monthly searches.