a tool for estimating the locations of users on twitter

This code is released under the 2-clause BSD license.

• python 2.7
• the python packages listed in reqs.pip
• Gisgraphy (http://www.gisgraphy.com/)
• data from Twitter's streaming API
• a Twitter account and application to access the REST API

You'll need to create a settings_dev.py or settings_prod.py in order to run localcrawl. The easiest way to do this is to copy settings_dev.py.template to settings_dev.py.

cp settings_dev.py.template settings_dev.py

You will also need to go through the three-legged oauth to get the information that goes in settings_dev.py . This will help you: http://benoitc.github.com/restkit/authentication.html#oauth

All the data you need to do location prediction is here: http://infolab.tamu.edu/static/users/jeff/friendloc_data.tgz

To get the crawling process started you will need tweets from Twitter's streaming API. I used @bde's https://github.com/bde/TwitterStreamSaver to collect them.

Once you have the gzipped tweets, you can pick the users to crawl with the following command:

gunzip -c ~/may/*/*.gz | ./gb.py -s mloc_users

The lo-calorie twitter neighborhood crawler.

This was started as a seperate project, but it is now contained in the localcrawl subdirectory. It is probably broken.

In the process of writing the FriendlyLocation paper, I renamed a few things. This code still uses the old names.

paper | code ++++++++++++++++++++++++++++++++++++++++++++++++ target users | mloc users home location | mloc, median_loc contact | neighbor median location error | mdist pContact | vect_ratios