“Why don’t you have any programmes about penguins?”: Subjectivity, measurement and personalisation.
As the world’s largest public sector broadcaster, we actually have lots of content we could show our audiences about penguins. However, deciding what a piece of content is about; whether that’s a TV programme, news article or radio broadcast, is a difficult and nuanced.
At a time when our organisation is focusing more than ever on using data to personalise experiences for individual users, the question of what our content is ‘about’ is becoming increasingly important. Without knowing what our content is about we can’t understand or measure what our audiences like.
But how do we decide this?
At the moment our editorial teams can describe content by making statements like ‘This programme is about penguins’ – but what does that mean? Is the whole programme about penguins? Does it actually just feature a penguin in one scene? How much do penguins have to be mentioned for the programme to really be about them?
Describing content is very subjective and what terms we should use to do this is something internal teams often come into conflict with each other over, this is before we’ve even considered how our audience may understand these things – what an audience member considers a programme to be ‘about’ varies with their context, expectations and needs.
Additionally, with the increasing use of data to automatically create audience experiences it’s possible that we are exacerbating this problem by glossing over the faceted, subjective nature of knowledge that, as humans, we understand implicitly.
So how do we resolve the tension between subjective understanding of content and data driven personalisation? How do we meaningfully measure how our audiences receive and interact with our content? And how do we do this responsibly, while fulfilling our public service remit to educate, entertain and inform?
This talk aims to provide some practical ways to address these problems by looking at: ▪ Clustering concepts using machine learning to gain better understanding of context and human in the loop machine learning. ▪ How and when to measure subjective things ▪ Structured content, and understanding content on a granular level for eg. individual scenes or paragraphs or segments.
Vicky is a UX Architect at the BBC, currently working on creating consistent navigation, and experience across the many tools journalists use to create content. Previously she worked at an agency helping charities organise all their complicated information into beautiful and positive user experiences, and even more previously she was a librarian. She lives in London and spends her spare time organising noisy music events and taking pictures of them.