Over The Air 2011: BBC Digital Public Space

Mo McRoberts - Developer, BBC Archive Development @nevali

Digital Public Space at Over The Air 2011

View more presentations from Mo McRoberts

• Trying to make BBC Digital Archive accessible
• Then working together with other organisations:
  
  • BFI, Kew, national maritime museum, royal opera house, british library, national archives, national library of scotland
  • have 25–30 organisations who have said yes to accessing data
  • but only have Mo to connect things!
• lots of separate catalogues
• each catalogue refers to things in asset stores
  
  • may not be able to get to asset stores, but linking catalogues by itself is useful
  • also link to external sources such as dbpedia, geonames, etc
• want to make the archives accessible to people other than archivists
• golden rule:
  
  • give everything a single, permanent URI
  • make the data about that thing accessible at that URI
• could try to fit everything into one giant, extensible XML schema
  
  • or else just go with RDF…
• can put all the RDF from catalogues into an RDF Aggregator
• wanted to find overlaps in the catalogues
• aggregator evaluates all info coming in and tries to find matches
  
  • not just exact matches, but close matches too
  • disambiguating is the hard part
• create lots of stub objects
  
  • people, places, events, things, …
• eventually want to have spindle in the hands of the public
• the archives themselves are slowly being digitised, but it takes quite a while
• BBC Redux captures and stores TV and radio, transcodes them and makes them available in various forms
  
  • been running since July 2007 for everything that’s been running centrally (not all local opt-outs)
• now has an API and developers’ guide
• available to developers for the duration of OverTheAir:
  
  • prototype RDF aggregator to query
  • API to Redux
  • references to Redux are not fully tested – may or may not work
• genome project:
  
  • scan in, OCR and codify all of the Radio Times issues
  • from 1920 to 2009
  • this and /programmes will provide a public API for all broadcasts ever
• three windows of content availability:
  
  1. free to air on iPlayer
  2. commercially useful
  3. out of commercial time: e.g. desert island discs (but no music)
• aiming to have content available in 10 years’ time
  
  • BBC Director General has committed to this
• 1 recent episode of Doctor Who has 80 rights clearances
  
  • an older episode would be worse as you would have to find the appropriate rights holders

Future of Mobile 08: Signposting on the New Paths of Discovery

Andrew Scott — Rummble

• Only 4.5% of your time is spent in a good GPS signal…
• CellID in city centres is good enough to allow you to track your movement along Oxford St
• 25% of flickr photos are now geotagged
• Under the Radar last week — a good proportion of companies had something to do with location, but they were spread throughout categories
• What went wrong with playtxt (Europe’s first location-based social network)?
  • Cost (on mobile)
  • Mobile usability
  • Location set was manual
  • Lack of public understanding
• What did Andrew learn from playtxt?
  • Privacy was not a barrier — less than 5% used privacy settings
  • No boundaries — went worldwide
  • 15x messages via SMS than by web
• “Who’s nearby?” is not a business — see loopt
  • US only launch
  • Restricted networks
  • Not useful enough — just text your friends!
  • Lots of competitors
• What is the business model?
  • Need to know not just who’s nearby, but what they’re doing — context of presence
• Current services
  • brightkite — iPhone app, location focus
  • limbo — focussed more around what you’re doing
  • whrrl — recommendations like Amazon
  • zkout — profile matching
• Differentiators for Rummmble
  • Instant; Personalised
  • Existing sites not enough:
  • Use trust networks rather than friend networks
  • Use similar ratings to expand relationships
  • Computationally expensive
  • Add in who you trust for what — using semantics & language taxonomies
  • Also computationally expensive
  • See linkeddata.org for sources of semantically linked data
  • Can use twine
    • though twine doesn’t look like it’s quite there yet, as with most semantic web tools…
  • Can import social graph rather than spamming all your friends
  • Has to be quick — within 45s
• Location detection is a commodity
  • Operators could scramble Cell IDs to make cell ID databases useless, but they would risk all their customers getting upset
  • An individual’s current location is also becoming a commodity