You are here

ADA: Creating journeys through the BBC Radio archive using linked data

The BBC has a wealth of permanently available programmes across a range of subjects as diverse as Vinegar, the A470, the Battle of Bosworth Field and Vanity. These programmes have valuable content but if they are not on air they are very hard to find. We wanted to create a route into these programmes that our audiences would find easy and intuitive to use, leading them on a journey of discovery.

Across the industry solutions used range from the heavily internally manually curated approach of Netflix , to the user-driven algorithmically determined approach of Spotify . We wanted to offer a more curated journey between programmes than we would have been able to do with an algorithm, with higher quality and consistency than if crowdsourced, but we had to balance this with the limited resource available for that curation effort.

Creating a hierarchy into which all of our programmes could be fitted was not viable: it would have taken far too long. Using an existing classification system would have still required customisation effort and we would have had no control over changes. Our user research told us that our users did not like being presented with the top categories of a hierarchical system as it required them to guess which category their chosen subject was filed under: they described this experience as being presented with a series of closed doors.

We came up with a system we call ADA: Automated Data Architecture. ADA does not require us to try to fit our content into a structure; rather we simply semantically mark up the content and let the structure reveal itself. By using linked data for semantic mark up our concepts were linked to others already modelled elsewhere in the semantic web. This meant that we were able to use concept linking from existing ontologies without having to create any ourselves, and also had the advantage that our concepts were automatically updated when changed elsewhere in the semantic web.

In my talk I’ll cover the way we've put linked data to use, where we've used editorial expertise rather than automation and how the combination of editorial effort and automated linking creates a serendipitous journey that could not have been achieved purely algorithmically.

Presentation Type: 
Talk
Language: 
English
Presenter: 
Presentation Audio: 
Audio Size: 
14.7
Presentation Visual: