December 03, 2003

Google Auto Stemming

Greg Notess reports that Google is automatically stemming words. Google Starts Auto Stemming Searches (Nov 28)

Google will look for some word variants automatically and bold the words in the snippets provided for each result. This is seen in the search -- socially responsible investing -- social, socially, investing, investments, investors, responsible, responsibility. You can turn it off by enclosing the whole phrase in quotation marks -- "socially responsible investing" but not for individual words.

It will do some singular / plural forms. Enter -- apple orange fruit salad -- and you get apples and oranges as well - even salads. But if you start with plural it won't look for singular - at least not for this salad.

While it is good that Google is starting to consider word variants, this unpredictablility is going to be annoying to searchers. When should they consider word variants and when not? How can they turn it off when they want precise results? Altavista's recognition of * as a truncation operator is a much better practice. Even Google's new ~ helped with word variants. But this arrangement of sometimes on and sometimes off will be confusing.

Note: Danny Sullivan found that + will stop the stemming. +investor (Dec 7)

Posted by Gwen at December 3, 2003 06:33 PM
Comments