±«Óãtv

COVID-19 Diary Study

The Internet & Society team have been processing the results of their research study into how people’s lives, and use of technology and media is changing during the COVID-19 crisis. The four-week study worked with three key groups: young people in low socio-economic groups, multi-generation households, and vulnerable isolating populations.

After interviewing participants they're analysing all the data, sorting and grouping it based on behaviour, motivation and emotions. Hannah is leading the data postcards analysis and Holly is working with UX&D design researchers to assess and plan the analysis.

±«Óãtv Together

As reported previously, ±«Óãtv Together is a new way to watch or listen together with remote friends and family, we developed in a collaboration between ±«Óãtv R&D, UX&D and the iPlayer team. We've now been able to learn a little about how its been used.

Since it became available in June it has been used around 30k times, predominantly by 16- 35 year-olds. Top shows that have users have shared recently include I May Destroy You, Normal People and Doctor Who.

Object Based Media

The Anansi team have been exploring StoryKit, a suite of tools for authoring and delivering programmes using Object-based Media developed by ±«Óãtv R&D's Future Experience Technologies section (FXT).

They've been exploring the potential relationship with the Orator, the authoring tool that IRFS built for interactive voice applications, which was used for applications such as The Unfortunates, and now used and extended by the ±«Óãtv Voice team.

Anansi are studying their findings and will be sharing their conclusions with FXT.

Data Team

The Data team has ended their sprint cycle with a reading week and a hackweek

Reading week allowed members of the team to pursue courses on AI and Deep Learning, and familiarise themselves with technologies such , and Kaldi's .

The hackweek was a focused on , which attempts to identify the sentiment directed at different entities and topics within text. This gives a more accurate understanding than analysing the sentiment of the text as a whole.

To make the week more challenging we analysed comments taken from social media as shown above. These tend to be less grammatical and have poorer spelling than professionally written text, which makes them trickier to process with standard Natural Language Processing tools.