Prototyping Weeknotes #80
As part of our housekeeping week, where we fix or tweak areas of our code and deployment infrastructure, Andrew and Dan have been creating a baseline virtual machine that we can use in future projects. This involved investigating loads of oddly named technologies such as Chef, Knife, Puppet and Vagrant. They managed to get a series of recipes that can be used to build machines locally or on production, using a stream-lined method for deploying applications using Git post-receive hooks, so setting up a local development version of a project should take just two commands. As part of this work, Duncan created and documented the creation of a multi-core Solr instance, and continued work on the web interface onto our internal Git repository.
Yves has been looking at expanding the language model he uses for automatic spech recognition with BBC tags, in order to not miss potential tags. He has been preparing for a number of ABC-IP-related sub-projects: an API for the automatic tagging of speech radio, and an algorithm for automated segmentation of radio programmes. Also, he has been preparing for the second RDF working group face-to-face, next week. Roderick has been looking into ways of visualising some of the data Yves generated for the ABC-IP project, and helped with the API to allow sharing with our project partners
Pete did a lot of work around conceptual designs for potential future development of the News Linking project. Sean's signed up for an upcoming Gstreamer conference, hoping to learn more about Gstreamer and DVB-T/S, as well as the project as a whole.
Olivier focused most of his week on a presentation and demo of our research and prototype around news and external linking to the product team for the BBC News website. With all the work done in the past few weeks in making the prototype functional and pretty much bug-free, and Pete's very exciting sketches showing how the ideas of the prototype could be put in practice, nothing could possibly go wrong....except the network did, right in the middle of the demo. The presentation went well, regardless, and we are now looking forward to showing the work and ideas to more colleagues on the editorial side of the BBC, who happen to be looking at the exact same questions at the moment.
Continuing on his bullet-point frenzy, Olivier then spent the end of the week preparing his talk for the upcoming Paris Web conference where he will be talking about the BBC and open linked data.
James Cridland, Secretary of the RadioDNS project came in to see some of our recent work and discuss RadioDNS, and then we had a beer or two. Kat wrote up the news research as part of the wrap up for "Following the News" and started scoping possible next phases of development. She finished writing a blog post about the recent RadioTAG trial - live soon.
Through the week, Chris G, Vicky, Tristan and I did a lot of prep for our 6 monthly workplan and team review, which we had on Friday with the General Manager for R&D, and the BBC's Chief Scientist. It went very well, and then it was the weekend.
This week, I loved this official picture of NPR's new boss, we were all glued to DQF news, and this post (ostensibly about long lamented - or hated, depending on your viewpoint - TV show Why Don't you) looking at scarcity, distribution and creation/ consumption, made us think.