BBC Research & Development is examining the results of an experiment investigating whether crowd-sourcing can help it make more effective use of metadata.
The Automatic Broadcast Content Interlinking Project (ABCIP) is a collaboration between BBC R&D and data specialist MetaBroadcast, and is part funded by the Technology Strategy Board.
In May, up to 2,000 participants from BBC online community Global Minds were asked to listen to short audio clips from the 70,000-programme World Service radio archive. They then identified the metadata tags they found relevant or erroneous in reference to the clips.
“Making archives available online is a costly business that typically involves high degrees of curation by skilled editorial staff,” explained BBC R&D programme manager Chris Godbert.
“As part of the ABC-IP project, we’re looking at whether we could publish large archives with less editorial effort by using computer processing and crowdsourcing techniques.”
The first prototype application to emerge from ABC-IP is Tellytopic, which merges programme information from BBC iPlayer, BBC Redux (an archive of broadcast material since mid-2007) and the World Service archive, to create a larger archive for users to browse. Tellytopic is built on top of Atlas, MetaBroadcast’s open-source TV and radio metadata platform, and combines automatically generated data with data sets about programmes assigned editorially by the BBC.
“Tellytopic exists for the duration of the ABC-IP project to help us learn about and develop methods of topic extraction for programmes and people [eg cast and crew] so that their metadata becomes a lot richer,” said MetaBroadcast chief operating officer and experience designer Mirona Iliescu.
“That allows for better linking, and an improved experience for media consumers in the case of EPGs built on Atlas.”