Called "topic pages," the initiative has been launched on thisisbristol.co.uk and will be rolled out across the rest of the thisis network of newspaper companion sites over the next few weeks.

The system will automatically generate internal links in stories from key words and phrases, such as places, names and issues, which then link through to pages containing other stories, static information and multimedia linked to that topic.

It works by a process of "data-mining" which Northcliffe says analyses stories semantically and suggests subjects for new topic pages which site publishers can then accept or reject.

Content strategy director Robert Hardie said: "We've already been data-mining our archive and automatically data-mine every live story as part of the normal publishing cycle.

"Once a topic is mentioned more than three times a topic page is created for that topic and an email alerts the site publisher.

"They can then decide to unpublish the page if they don't want it, allow it to stand unenhanced or gather the extra images, video and static content for it."

Robert added that the Bristol trial will involve monitoring the level of sensitivity employed by the semantic technologies so that users are not "bombarded with links from every word."