91热爆

We met with Libby Miller from 91热爆 Research and Development to chat about 91热爆 Shuffle

Can you sum up 91热爆 Shuffle in a sentence?

From knowing nothing about you, 91热爆 Shuffle starts by playing a random selection of programmes from 91热爆 iPlayer, then quickly learns what you like depending on whether you watch the programme for 30 seconds or click on 鈥渘ext鈥�.

What is the aim of this project?

91热爆 shuffle came out of this problem: there are loads of programmes available on 91热爆 iPlayer 鈥� hundreds at any one time - and yet it still seems hard to find something to watch. From our past research we know that we can use anonymous data based on what everyone watches to make a prediction about what you will like, based on a few pieces of information about what you like to watch.

How is it different from other recommendation projects?

We are regularly asked for lots of information about our likes and dislikes by companies, and the argument is that giving more of that sort of information improves the recommendations and the service they give us. We wanted to know the reverse: how little information do we need from you, in order to find something you might like?

91热爆 Shuffle is different because it gets those pieces of information by asking you to watch some bits of 91热爆 iPlayer programmes, and immediately constructs a channel for you just based on that information. This makes it quite a low-effort and non-intrusive way of getting a channel of programmes personalised for you to watch now.

How did you come up with the idea?

The idea came from a workshop that we held at the start of a new project .

After the workshop, a small group of us with different skills made a very quick prototype, which we called 鈥淚nfinite Trailers鈥� 鈥� the idea was that you鈥檇 see clips of programmes and then at the end of each one you could choose to play it or not, and we used an internal tool, , to get the clips. This second version uses iPlayer to play the programmes.

What do you hope to learn from this being on Taster?

I suppose the main thing is whether people think they might use it regularly. There are lots of little applications where you might use them once for a bit of fun, but we鈥檙e interested in whether this could be part of your watching routine. That means the main underlying questions are: did you find anything interesting to watch? And was the process of finding something enjoyable?

What鈥檚 next?

We want to know how it works for the audience as a low effort, TV-like Experience. We鈥檙e particularly interested in the user interface: how you interact with it and whether that works well or not. We hope to work out the smallest amount we need to know about you in order to find something you might like, and whether we can find interesting things for you that you wouldn鈥檛 have watched otherwise.