91热爆

Tuesday was Vicky's last day in the Lab, very sad. Theo recruited some people to test the current version of Programme List (nee Watch Later). Some really good steers for immediate improvement and functionalities for 听the longer term. Very good response 鈥� we quote one participant - 鈥渋t鈥檚 like someone has been in my head a created something I need鈥� 听Olivier's been documenting the news linking project on a day-to-day basis. No more hacking SQL databases him this week, as both Chris N and Duncan have taken over in this area. Olivier and Jo turn the tables by conducting an interview on news and linking with the editor of the magazine section of 91热爆 News interactive. Great interview, and Olivier thinks it's quite exciting (and a little ironic) to interview a journalist. I go to visit the new R&D Lab in Media City - very nice.

On Wednesday we continued our team sessions on News Companion ideas generation. A number of stages passed through to get all the implicate information on the board and understood by everyone stakeholders & their requirements/interests 鈥� mapping to R&D鈥檚 research objectives. We also created some definitions of key terms 鈥� all of which turned out to have several meanings.. We began a series of idea iterations 听to resolve working themes key themes 鈥� based on News research findings and then evolved to become meaningful to us. Kat and Tristan meet an ex-colleague to talk about future radio ideas. Sean has a good chat with Justin Kand Matt P about the R&D systems infrastructure, and how it could be improved, in particular by introducing cloud provision. Olivier worked with Yves R (soon to join our Lab full time) on an updated presentation on linked data and the 91热爆, to be given to our colleagues in News and Knowledge.听

On Thursday, Tristan has a couple of chats with people about potential uses of some of our past prototypes. Tristan sits in on some user testing sessions of the Programme List that Theo is running with some colleagues from the office across the corridor from ours. We wanted some fresh eyes on it and it was broadly positive but with some (now) obvious tweaks that we should do. From this we generate some more tasks, then review the outstanding tasks and their priorities. Tristan has done some thinking about user-owned data, reworking and tightening our argument. We should be ready to start presenting this in a couple of weeks. Kat spends a lot of time with A&M on a RadioVIS service for Radio 1 and 1Xtra, due to go live soon. 听Olivier helped Duncan R (another R&D colleague), with his first steps as our latest rep to a W3C group.

Chris L spent a lot of time during the week continuing with the News Statistics work, looking mainly at implementing a document-similarity library so that we can do visualisations and clustering similar to this technique presented by Jonathan Stray:

To that end Chris 听wrote (and released) an open-source Ruby library for doing the calculation:() and this produced this example image for keyword clustering

听

听

听

听

听

听

听

听

听

听

听

听

听

Things we liked this week

听

• 听(a cross-language data serialization library with built-in RPC. Sean thinks he'll be using this in his own future messaging work.)
• (Amazon release a comprehensive Ruby SDK for Amazon Web Services)
• . (A good review paper on document clustering methods.)

听

(Tardy update from me this week. Apologies..)

听

听