Logo

Logo

Google is improving 10% search results by understanding natural speech queries

The tech giant claims that it has improved its ability to analyse search queries and offer relevant results for as many as one in ten queries

Google is improving 10% search results by understanding natural speech queries

Representational image (Photo: iStock)

Google has announced a change to its core search algorithms that it claims can better understand conversational queries and throw more relevant results.

By applying improved natural language analysis, the tech giant claims that it has improved its ability to analyse search queries and offer relevant results for as many as one in ten queries in the US English and support for other countries will come later.

Advertisement

In essence, Google is claiming that it is working on improving the search results by making the system better understand how each word in a sentence is related to others. Previously, the search algorithm treated a sentence as a bag of words. Therefore, it looked at the important words like pharmacy, or medicines, and simply returned local results. The new algorithms are now able to understand the context of the words like ‘for’ and ‘to’ to realise what the query actually mean.

Advertisement

The technology behind the new neural network is called “Bidirectional Encoder Representations from Transformers” (BERT), which Google first introduced last year.

“By applying BERT models to both rankings and featured snippets in search, we’re able to do a much better job helping you find useful information,” Pandu Nayak, Google Fellow and Vice President, Search, said in a blog post on Friday.

On average, Google sees billions of searches every day, and 15 per cent of those queries are ones it has not seen before.

BERT models can consider the context of a word by looking at the words that come before and after it—particularly useful for understanding the intent behind search queries.

“Some of the models we can build with BERT are so complex that they push the limits of what we can do using traditional hardware, so for the first time we’re using the latest Cloud TPUs (Tensor Processing Unit) to serve search results and get you more relevant information quickly,” said Nayak.

(With input from agencies)

Advertisement