News algorithms and how stories age

News is about what’s current, but different stories age at different rates.

Some stories are relevant for weeks (such as Metro’s Ebola facts article). Some breaking stories are old within hours.

At Metro we’ve been automating how stories drop down the page. Currently, there’s a “one site fits all” approach. It’s good, but it could be better…

What if we let users tell us when a story gets old? Falling clicks means older. This could be a way to get the most out of each story.

Unfortunately, it’s hard to know when clicks are falling because clicks are also dependent on where a story appears on a page. There is a way to take this into account though – we can find the average number of clicks for a each position and weight results accordingly.

I’ve built a demo using Redis/Play. (It extends our real time app).

For some of the gory details see my older post.


Leave a Reply

%d bloggers like this: