Close window

A natural phenomenon

The Met Office is using natural language interfaces to create new ways of delivering weather forecasts.

From newspapers to radio, from television to smartphones, the Met Office has always embraced the latest media when delivering forecasts.

Now our developers are focusing on natural language interfaces to deliver forecasts via Amazon Alexa and a ‘weatherbot’ on the Met Office Facebook page.

“Natural language interfaces let you interact with a system with humanistic speech or text,” explains Thomas Powell, Met Office Informatics Lab Technologist. “So you can ask literal questions and receive an understandable answer.”

The technology was developed as a pilot by Thomas and his colleague Niall Robinson. The Met Office Informatics Lab worked closely with Amazon to include a weather update as part of the daily briefing that Alexa delivers to users in the home. Similar technology sits behind the weatherbot, which enables users to ask questions like, “Do I need to take an umbrella?”

Tomorrow’s world

Although devices like Alexa aren’t yet in every home, Chris Frost, Head of Digital at the Met Office, believes there will be a huge uptake in the technology.

“As children grow up with devices like Alexa, it will be second nature for them to control objects by talking to them,” says Chris. “Like telling the light bulbs to dim or turning the TV on or off.”

The Met Office is determined to be at the forefront of the latest developments, making weather forecasting available to people however they wish to access it.

As Chris says, it delivers directly to their remit of working together for protection, prosperity and wellbeing: “We’re saying the right thing at the right time, to the right audience, so that people can take action to stay safe.”

A question of working smarter

Natural language interfaces could transform the way we work at the Met Office. Our systems hold a vast amount of data, so accessing and interrogating it can be a complex process. Scientists have to become familiar with data language and code, and request information from colleagues.

So imagine if someone can call up data simply by asking a natural question like “What’s the rainfall data for Beijing this year?”

“If someone comes straight from a physics degree, they’re not going to know all of our internal storage systems,” explains Thomas Powell. “But if they can get the data they need and concentrate on the science part, it could really free them up to focus on what they do.”

To take another example, journalists often pose the Met Office questions like, “Is this the wettest day of the year?” With a natural language interface in place, the engine could crunch the numbers and come back with an answer in superfast time. It could change the way all of us work.