Things that are like the things you like

How on earth are you meant to keep track of everything happening these days?

The Internet! It’s just getting bigger by the second, isn’t it? So here at Wavii we’re doing our best to not only distill the news we find into bite size chunks but to help you find the bites you want.

A key aspect of helping you discover content is through recommendations, and one of the fundamental technologies we use to do this is the collaborative filter.

The main idea behind a collaborative filter is that we think you’ll like things that are like the things you like.

Breaking this idea into two parts…

We think you’ll like the things that are like the things you like.

How do we know what you like? Well, we decide this based on your interaction with the site including, primarily, the topics you already follow.

We think you’ll like the things that are like the things you like.

How do we know what things are like each other? This is where the art of recommendation systems comes into play; what “signals” can we use to find topics that are like each other? Once we have them, how can we combine them?

Huh? How?

For example, one of the many signals we use at Wavii to denote similarity between topics is the idea of bilateral links on Wikipedia.

Consider the Wikipedia page for 90s grunge sensation Pearl Jam. In their infobox (that summary box on the right hand side) you can see there’s a link to their lead singer, Eddie Vedder.

If we then go look at the page for Eddie Vedder we can see there’s a link back to Pearl Jam.

We use this bilateral link as a signal to say that Pearl Jam and Eddie Vedder have some sort of relationship. If we consider a whole bunch of these links we can build reasonably interesting graphs:

Click through to view a bigger version, can you spot the great Kardashian constellation?

The challenge lies in how we combine this signal with the numerous others we have and we’re working hard on making it perfect! If you feel like having a hack yourself here’s the link data.

About Mat

Data nerd at Wavii
This entry was posted in Engineering. Bookmark the permalink.

Share your thoughts:

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s