How Google Applies Science to Search (Part 1)

Some notes from Dr. Craig Nevill-Manning’s presentation at this year’s Webstock conference, as compiled by Kalena Jordan.

It’s quite a fascinating “peep under the hood” of the many and varied components of Google search, including ensuring the spellings of search terms are correct, and in the right context:

As the popularity of the search engine grew, Google needed to be able to spell-correct lots of obscure words. So his solution was to take a sampling of content from the entire web. Craig’s team came up with a algorithmic model and ran it over the web. He discovered that there were several correct answers to the same question. For example, words like “kofee” could mean either the searcher is seeking a cup of java or information about Kofee/Kofi Anan.

Via Webstock.