In an attempt to provide ever more meaningful answers to users’ search queries, Google has now launched a series of new algorithms that is designed to compress long and complicated sentences into easy-to-read snippets.
The sentence compression algorithms are now live and are currently employed in Google’s desktop search results. The technology is based on advances in machine learning and the main purpose of the software is to take out the most useful chunks of meaningful information from a long sentence or paragraph before returning a more concise and compressed version of that information in response to a unique query.
So, How Do Google Actually Make The Algorithms To Work?
In order to get Google to “understand” and “respond” to a search query before returning a useful response, the developers used thousands of old news paper articles which teach the machine to see how longer articles can be compressed into headlines which summarize the information that follows.
The team at Orr, who are helping to develop these algorithms, also rely on a diverse team of PhD linguistic experts who are able to demonstrate sentence compression techniques whilst labelling components of modern speech in a way that the neural network software finds useful in a global effort to teach the virtual brain how we piece complicated language together.
With more and more apps and search assistants relying on rich featured snippets to provide their users with useful information, it is hardly surprising to find Google spearheading these advances.