This module can be used to generate datasets that aggregate geographical information about editors. The source of the data is the recentchanges table of the Mediawiki MySql databases. Currently two datasets are generated

• Each row shows the number of editors of certain activity level for a given country. The files are tab seperated, there is one file per Wikipedia project (e.g. enwp).

  Country, total editors, total active editors (5+), total very active editors (100+)

• Each row shows a country and total number of edits, followed by a list of the top ten cities and the percentage of edits made in that city. The files are tab seperated, there is one file per Wikipedia project (e.g. enwp).

  Country, total edits, [city, {0.0-10.0} weight compared to largest contributor city]

One needs access to IP addresses to create geo coded datasets from wikipedia. Wikipedia's privacy policy states that IP addresses are only stored for a limited period of time. The datasets generated by this module do not contain information about indidual editors, all datapoints are aggregated on a city or country level.

• Python > 2.6
• argparse
• mySQLdb
• mysql_config
• geoip
• GeoIP City Database from Mindmind

In 'process_data.py', set the following directories.

• output : generated geo coded data files

Configure access to the mysql databases by configuring the mysql_config.py file. The login info has to to be configured by creating the file ~/.my.cnf with the following conent:

[client]
user = USERNAME
password = PASSWORD

The data is retrieved using a server-side mySQLdb cursor. The tables queried are:

• checkuser table, cu_changes (Main)
• recentchanges table (Alternative)
• user_groups table. For bot classification.

No joins are performed.

Point geo_coding.geoIP_fn to the GeoIP City Database.

Note: Any files that already exist in the cofingured data/output directories will be overwritten. None of the already existing files will be deleted. At the moment no date-specific information is included anywhere in the files or the file names, it is best to run the script with empty directories.

Simply run:

python process_data.py

• Add date specific information in the data files and the file names
• Tests!
• Create a package