IRFS Weeknotes #105
This week the team welcomed visitors from , a Norwegian research institute located far into the Arctic circle. They were partners with us on the P2P-Next project and came to present their work on multi-device synchronisation of media presentations. Chris Needham and Dominic ran a couple of workshop sessions to generate ideas and identify areas of future work. Our guests also gave a lunchtime lecture (the first one hosted from the Central London lab) on this topic.
Duncan found time this week to migrate Whereabouts to . Whereabouts is a small, internal web app we wrote a while ago using to replace the deluge of emails sent around the team listing where everyone's going to be during the week. Instead, users go to the app, fill in their locations and it's done.
The Whereabouts system shows the team's location for a single week
We're hoping to use this side-project as a test-bed for trying out new technologies and also for building interesting inputs (emails, text messages, mobile apps) and outputs (receipt printers, TVs in the office) on the API. It chimed well with Zach Holman's about how hacking on side-projects helps to build a team culture and fosters collaboration.
Barbara has been working with all the FI-Content project partners to work out logistics for project review at the Commission, including agreeing the presentation material and what demos to show. First review comments from partners were very complimentary on the massive use case document she's been putting together with Dominic. There's a bit of refinement to be done but she's pretty confident that this deliverable will impress the EU Commissioner!
Staying with FI-Content, there have been discussions to define the essential features and user journeys for the Chrome extension prototype we're working on. Joanne's been planning the next evaluation of the project: working out research questions and recruiting participants. Dan's been building the extension and Pete's been designing the in-browser interface elements to support our current investigations into viewing activity and privacy.
Pete's also developing a presentation to best communicate the historical context, demonstrations and explorations of the Audio API project while Chris L worked on the documentation for the project prototypes. Chris then jetted off to "sunny Budapest" to attend and give a talk at (it seemed to go down well). We're looking forward to welcoming Chris back to the lab where we've been getting a great view of the downpour over London.
Yves wrote an enormous blog post about the WWW 2012 conference last week on the R&D blog. As a result of his presentations at the conference he's been involved in discussions with various institutions which could lead to some collaborations. He's also been making some progress on programme segmentation and contributor identification.
Tristan's also been writing this week, but probably not in the same voluminous amounts as Yves. He's been writing about what we do in various forms, sketching the world of identity and personal data in the Ö÷²¥´óÐã. He's been thinking and talking about making better ways to turn our ideas into good projects: 10% time, improving the visibility of ideas and incoming projects, structuring the first exploration sprint of a project.
Tristan also learnt how to demo RadioTag. Luckily George was around to get our prototype radio working!
Vicky's working on a proposal for a user study about authentication (people & devices), behaviours and expectations. She spent her 10% time on a "roar to explore" demonstrator to package up the work done on an audio classifier so that it's easy to demo and test with users. She's also talking to people about an "Internet of Things North" event, possibly in June.
Libby got the NoTube wrapup blogpost out, and been looking at the implementation of RadioVIS in a bit more detail. She also met with Sean, Theo and Chris Newell about the application of the egBox to VistaTV and reports that: "It looks great and I'm very keen to be a guinea pig for trying it out."
Chris Newell has been refining and documenting his Javascript recommender module which provides client-side recommendations. The current version uses a Weighted algorithm which can easily accommodate positive and negative feedback. The algorithm requires a correlation matrix describing the similarity between each pair of programmes which can either be based on the number of common viewers (collaborative filtering) or the number of common attributes (metadata-based filtering). A model based on this matrix is rebuilt every hour by the server and delivered to clients along with the recommender algorithm code.
Finally, Dan and I gave a quick overview at our weekly show and tell of the most interesting presentations from that we both attended.
As well as the to speakers we made a collection of some of our favourite things:
Comments Post your comment