Google Explains Search Is Better At Understanding Meaning

November 16, 2015 - Written By John Anon

If you are looking for information on something, anything in fact, chances are your first port of call is Google. After all, that is why the popular phrase ‘Google it’ exists. Google does seem to be considered the be-all-and-end-all when it comes to information and with good reason. It does know a lot. However, it does not know everything. But it is learning and this is largely thanks to what Google refers to as its Knowledge Graph. In fact, according to a blog post sent out this morning by Google, the Search is not just learning, but starting to understand meaning too.

The blog post sent out today by Google looks to explain that as Search has evolved, it is finally starting understanding the meaning behind some of the questions it is being asked. More specifically, the intention behind the various questions. It is thought that the idea is to make Search more capable of understanding a question much more like a person would, then say, a machine. In fact, it was only last week when reports emerged on how Google are investing in artificial intelligence measures to try and help Search to learn to improve itself. Now when you ask your Android device a question (or of course, type it), the Google app will be able to understand more complex aspects like superlatives “What are the largest cities in Texas?“, ordered items and even date relevant queries like “What songs did Taylor Swift record in 2014?

Interestingly, Google does conclude the blog posting by mentioning where Search is still weak at understanding questions and it seems to specifically relate to where a question has too many possible meanings. The example Google give for the weakness is when you ask Google a question like “Who was Dakota Johnson’s mom in the movie?“. In this instance, Google notes that Search will interpret this question in relation to Dakota Johnson’s real mom, Melanie Griffith and hence provide her movies instead of the target of the question, Jennifer Ehle. Of course, it probably should be expected that Google is working on finding a solution to this sort of more complex questioning in the future.