largest broadcaster in the world by number of employees, with over 20,000 staff in total. Just in the UK - 9 TV Channels, 56 radio stations (10 national radio stations, 6 country-wide radio stations and 40 local radio stations, plus world service.)
Launched in 1997, the website was hand coded, and all the links manually checked.
It’s now updated by means of content management and automated content publishing systems.
We publish programme schedules automatically through playout systems, so every programme has a page, they are not always easy to find.
The news app allow people to follow a single subject.
Our music pages are also dynamically populated, content from Wikipedia and Musicbrainz populates the page, while any content we create that is tagged with the artist name will be added to the page.
Published openly, free to use.
Although, news, sport and music are now being published dynamically, there’s still no automation for the rest of the programmes Slide 3. The difference with programmes Structure Sport is already structured (teams, leagues,etc) Parliament/constituencies likewise News had editorial structure in the first place, Business > Companies, Economy, etc Some programmes have a hierarchy, but inconsistent, K&M , IOT, DID comparison. Relevance: date of publishing much less relevant with study topics.
• “We decided to see if we could user the user-created content of Wikipedia to create automated navigation…” • Wikipedia holds rich – and despite its reputation, accurate! – info on almost every subject • NEXT SLIDE ZOOMS IN ON CATEGORIES
• “The principle is quite simple, each programme is about a subject and that subject already has categories attached in Wikipedia…” • Since the categories are entered by people according to what they think is important, they are as editorially valuable as crowdsourcing, without the effort and inconsistency of doing this ourselves over a large dataset.
• “Using the Wikipedia tags we were able to construct queries to extract all of the category data and check how many categories matched across different programmes...” • “The whole process only takes a couple of minutes per programme and gives and average of 7 categories per programme…” • “540 matching categories were found in total…” • Some intriguing links and connections can be found – Connections that sometimes aren’t immediately obvious but are extremely insightful
• “Starting with the In Our Time archive because it has a broad range of subjects…” • In Our Time is perfectly suited to experiments around categories– Such a broad range of topics • The existing archive is manually categorised with a small set of very high-level groupings
• Automatically create and maintain these new connections – Programmatically
Jo Kent | ADA – Opening up the BBC archive with linked data
Introducing ADA: Automated
Radio & Music Interactive
• The British Broadcasting
Corporation (BBC) is the UK's public-
• It is the world's oldest national
broadcasting organisation and the
largest broadcaster in the world.
• It produces news articles, television
and radio programmes 24 hours a
• Somehow that content needs to be
The BBC in 1997
Access to our ontologies and objects
Our programmes archive offer
• Only brand
• Expensive to
What about the rest of our programmes?
for all of
But do we really want a hierarchy
Just as we no longer
use directories for the
web, we need to look
beyond hierarchies for
Homophily and user experience
Homophily (i.e., love of the same) – people like things
that represent their worldview.
The more links you present, the greater the trust in your
1. Ilan Lobel and Evan Sadler. Preferences, Homophily, and Social Learning. In
Operations Research, 2014
ADA• Feedback has been positive and constructive
Just a quick look delighted me … I can find all sorts of stuff I never knew existed
The new beta categories database is SUPERB!! Congratulations to the programmer. Truly
wonderful. Many, many thanks. Any chance applying it to other science-type podcasts ?
Really ingenious – like it a lot, suggests connections, stimulates interest
Thanks for arranging the categories of In Our Time. I would absolutely love TV programs to be
this accessible. Not ones chosen by experts or highlights of, but all of whole series arranged so
that one can follow ones own interests.
This is a marvellous development! Google can be rewarding, but your development plugs us
into a more manageable field of knowledge – one that’s nevertheless wide ranging and high
quality. I love the joining-up.
IOT's beta site is simply the single best I've experienced. Now that I have
experienced it, I don't want to use the current site's format.
• A single subject input automatically generates a wide range of categories
• No need for producers to think about categories/genres
Subject input ADA service processing Categories generated Connections created