Software automates news sites' geotagging

News organisations have been experimenting with geographic localisation software that has received funding from CIA and has been used by the oil industry to develop highly personalised local news products.

Massachusetts-based software company MetaCarta says a large UK news organisation is about to announce that it will begin using its artificial intelligence software to attach geographical coordinates to online news stories.

This process, known as geotagging or geocoding, is essential to developing localised news sites, a strategy that several regional news organisations in the UK – most notably Archant – plan to implement this year.

Reporters or photographers can manually geotag news stories using portable GPS receivers while on assignment, or by cross-referencing stories with coordinates gleaned from online tools like Google Earth. But automating the process could be essential to making geotagging consistent enough to be useful.

‘In the many discussions we’ve had with news organisations, many have said that they just want their reporters to geocode their articles,’says MetaCarta director of content services, Rick Hutton.

‘But reporters aren’t necessarily able to do that very well. A reporter will typically only geocode it to the main place that an article is about. If there are 10 places mentioned, we can geocode all of them and do it on the fly.”

Natural language processing

Instead of manual coding, Metacarta uses linguistic software techniques known as ‘natural language processing’ to identify geographical terms mentioned in a story’s copy and cross-references it to a gazetteer – a list of more than 11 million place-names and their correspondent geographical coordinates.

The gazetteer is customised for each news orgnaisation that wants to use the technology. Working with the British news organisation, for example, MetaCarta’s software has had to learn to distinguish between the British and American university towns called Cambridge.

‘I believe there are about 85 Parises in the world. But we have the ability to identify the right one – and in fact that it’s not Paris Hilton but rather Paris, Texas or Paris, France,’he says.

Geographic search

Linking information with geography creates new opportunities for retrieving the most relevant information, explains Hutton: ‘If you wanted to find out about a murder in Sudbury with a keyword search engine, you would only get those documents that mention the word ‘Sudbury’. With our technology, it would find anything that happened in the area around Sudbury.”

‘The document might mention a village, or a street in Sudbury, or even one particular store, but it might not ever mention the name of the town itself.”

Founded by graduates of the MIT in 2000, MetaCarta found many of its early customers in the American public sector – from the US intelligence agencies to the Smithsonian Institution.

Petroleum companies such as Shell have also used the software to organise millions of internal documents about places and areas where they have mining interests. But now, Metacarta is marketing the product to online publishers.

Targeted e-mail news alerts

In Texas, the San Antonio Express-News uses the software on its website, MySA.com, to send customised e-mail alerts whenever it publishes a local-interest story. Several other news organisations are about to launch similar products based on MetaCarta Local Alerts, Hutton told Press Gazette.

‘We’re working with a rather large news provider in the UK and two more newspapers in the United States,’he said.

Reuters, meanwhile, uses MetaCarta NewsMap to plot stories on its US website to an online map. The same product is also used by the startup website YourStreet to aggregate news stories from other sources on a Google Map.

Comments
No comments to display

Leave a Reply

Your email address will not be published. Required fields are marked *

five + 12 =

CLOSE
CLOSE