Close menu

Research & Development

Posted by Barbara Zambrini on , last updated

These are weekly notes from the Internet Research & Future Services team in BBC R&D where we share what we do. We work in the open, using technology and design to make new things on the internet. You can follow us on Twitter at @bbcirfs

One thing I've learned since joining the IRFS team is that everybody likes variety. No week is the same as the previous one and there is always somebody doing exciting stuff.

This excludes writing periodic reports on an EU project as I did. But even I managed to step away from that long document to catch up with few of us who are looking at possibilities for prototyping an enhanced radio experience.

In Salford when not planning the roadmap for IoT work and preparing presentations, Vicky and Jasmine have been soldering the circuitry for the radio fob prototype and getting an initial set-up of a raspberry pi and clone-able SD card OS in preparation for prototyping workshops. Libby has also joined them in Salford helping them run a workshop with Childrens, Mobile, TV platforms and Marketing &Audiences to explore the set top box for kids idea. It emerged that kids as creators, not just consumers, could be something that differentiate the BBC from other organisations. Back in London, Libby ran another workshop with the majority of the team planning our work in the connected TV space for the FI-Content phase 2.

Trialling a new approach for engineers working on several projects in parallel, Chris L. and others started a two-week sprint working on the VistaTV dashboard and the World Service prototype. He tweaked the search interface for the WS prototype and added a mechanism to invite people on the waiting list. Together with Dan he's been building the backend architecture for the VistaTV dashboard using Elasticsearch as a datastore and a couple of light-weight eventmachine servers to get the data out to the client application. While Andrew N. started looking at the existing Radio Dashboard demo code for VistaTV converting the existing "bubble chart" to use the excellent d3 data visualisation library.

James and Yves have been setting up a full Kiwi (speech-to-text + tagging) pipeline for Snippets using Amazon Web Services and started by processing all Radio 4 content. After a bit of a bumpy start it's working great! He was also working on the Snippets framestore adding support for retrieving images from iPlayer's image services for pips episodes IDs.

When not busy on the Snippets work, Yves has been doing some investigations related to automated tagging of news articles, trying out a new algorithm using the SLEPc library and very large Wikipedia/DBpedia dumps. He was also working on a paper with Jana for a WIAMIS workshop next summer, on recent work on automatic analysis in BBC R&D.

After battling with some heavy snow, Matt has been writing a little java app to interrogate our Lucene/Solr indexes allowing him to find out the IDs of all the documents in Solr in a matter of minutes instead than hours.

Chris Needham started writing a blog post on the results from the authentication project while continuing to work on improving the egBox code and documentation. Recently he's been using PhantomJS to test our client-side CoffeeScript code.

While Denise successfully implemented Jana’s Machine Learning Framework with supplied audio examples, Sam has been analysing the mood of 9 months worth of subtitles finding that Sunday mornings on BBC One are the least exciting and Monday mornings are the most positive :-)

Chris Newell has added IP address based location information to our anonymised viewing data streams. This will allow the VistaTV stream processing engine to do a real-time analysis of viewing behaviour by user location. The next step will be to resolve the individual UK region codes into their corresponding BBC TV regions.

Tristan has been writing some copy to clarify some features of the World Service prototype and had a conversation with Jake Berger from the Digtial Public Space project and Simon Popple from the Institute of Communication Studies at University of Leeds. Simon has been working with the BBC on archives and audiences and is interested in how you get people to engage with, and create with, archives and how they can help foster mutual understanding between people.

Finally, Olivier's week could be summarised in a word - testing. He has been dealing with a specific bug found on the beta R&D website and doing conformance testing on the W3C Audio spec with Chris Lowis and Matt. But aside from testing, he also had discussions on synchronisation and "multiple screens" scenarios for W3C Web and TV.

Some interesting links: