Inspired by Tim O’Reilly’s dictum
“Take the intelligence of all your users and put it in the interface“,
I’ve asked my colleague Nataly to adapt Amazon’s “Customers who bought” feature for the Chalkface site as “Teachers who bought…”
This is what’s known as an ‘implicit recommendation engine’. The theory is that by looking at what other teachers think (i.e. tapping their intelligence), you will be more quickly guided to resources that are likely to be useful to you.
The base data we’ve used is actually the sample downloads log. This is a more reliable indicator of real teacher intentions, because unlike purchase data, the teacher-generated data are not diluted by orders placed by school administrators who may be ordering on behalf of more than one person.
We only show the top four linkages. Browsing through them, I’m struck by how tight the clustering is. I’d like to find a way to say “a significant minority of teachers also bought this…” as a way of pointing you towards undiscovered gems that may also be of value to you. Not sure how to represent that algorithmically, though.
Here’s an example. See what you think.
Yacapaca 