Google recent announced an update to its search engine called Hummingbird. The change was touted by Google as one of the largest changes ever made to its system, yet searchers hardly any impact on rankings. This was because Hummingbird wasn’t an algorithm change; rather it was an architecture change that lays the groundwork for many future changes. Today we dive into the patent behind that change, as we hope to unravel future updates before they are even announced.
The patent behind Hummingbird is titled, “Determining the Meaning of a Knowledge Item using Document Based Information”. The idea behind the patent is that search results can be improved if Google can understand the meaning or intent behind a query, as opposed to strictly trying to match keywords to a webpage.
To make this intent based search possible, Google must first compile a giant collection of people, places, things and concepts. Each item in that collection is a unit of knowledge which likely relates to other items in the collection. For example, the item, “car” is related in some way to the item “gasoline” which is related to the item “fire”. By understanding these relationships, Google can begin to think about how the words used in a query relate to each other, giving clues as to what information the user is really looking for.
Understanding this new world of word relationships should prompt us to change the way we think about keyword research and even SEO. Check out the recording here: