Eigenlayouts

An investigation of web layouts through image data. Pulls down website homepage via PhantomJS for later processing.

Requirements

To retrieve the images, you'll first need to download a corpus of domains. A good corpus to start from would be the list of top sites from Alexa.

To run, seed your list of domains as a directory full of dummy files with

make mkdomains

Once that finishes, you should run:

make -i

to start pulling down domain data serially. To run multiple jobs at one time, run:

make -i -jN

where N is the number of jobs run at time.