OK, computer: Automated news to grow

Financial news and data produced by computers to be read by computers looks set to become an increasing part of agency journalism, judging by a new study released by Thomson Reuters.

The survey has found that financial firms using advanced algorithmic trading techniques expect IT-driven news analysis tools to substantially shorten the time it takes them to take advantage of new information.

Ventanta Research spoke to 113 specialist financial professionals and found that two-thirds expect increased automation in the analysis of news.

Over the past two years, before their merger, both Thomson and Reuters have been working on several products which deliver data and other news to its financial services customers in a format that can be read and acted on automatically by computers. Much of this material is also created automatically.

'It's still the same news – we're not really providing a different piece of news, we're just providing it in a different format,'said James Chenery, Thomson Reuters' business development manager for quantitative and event-driven trading products.

'We're providing particular pieces of data, which used to be very much textual in their delivery, as a much more structured message whether its XML or Reuters' realtime for the market feed."

To aid the creation of these news feeds, Reuters has been developing tools to automate the process of extracting machine-readable data from standard narrative text and to embed semantic metadata – machine-readable information about the tone of stories. These include services that automatically extract key figures from releases of economic information, such as official statistics and company earnings releases, and to rapidly provide that data in a format that can be included in client firms' trading algorithms.

'Earnings releases are primarily numerical in nature, but when they are covered by text – a sentence – then it becomes more difficult for a machine to understand and comprehend that information,'Chenery explained.

This year, Reuters started adding machine-interpretable semantic data to otherwise conventional, human-readable news stories.

Thomson Reuters expects machine-readable news to become an increasingly important part increasingly important part of the company's business and has said that the automated extraction and delivery of data used by these tools will free journalists to do other, less mechanical work.

Chenery said: 'By providing that kind of automation and technology, it allows the journalists to spend more time developing exclusives or writing up more information.

'We don't want a relatively highly paid journalist putting a textual number into some sort of system to send it out when we can do that automatically and probably faster and would rather have that journalist writing an analysis piece or something like that.'

Sign up for our free weekly digital magazine, Press Gazette Journalism Weekly, and daily newsletter
To contact Press Gazette with a story call 0207 936 6433
or email pged@pressgazette.co.uk
To advertise, please call 0207 936 6764.