Andy Halterman
Andy Halterman
I'm embarking on an overhaul of Mordecai to improve performance and address many of the issue here. If you use the library, please help out by answering 2 minutes of...
Right now, Mordecai calculates features like the number of results per country or the result with the most alternative names from the first 50 results that come back from the...
srsly
After a long delay, a new version of Mordecai is out: https://github.com/ahalterman/mordecai3 The new version is a complete rewrite from the ground up and has the following changes: - uses...
Make a setting for new_verb_length with equivalent functionality to the new_actor_length. Basically, pull out verbs even if they aren't in the dictionaries. This will allow us to rapidly harvest new...
PETRARCH comes bundled with the dictionaries it needs to run. These dictionaries have their own repo and change frequently. What's the best way to keep people using the most up-to-date...
Right now, the geocoding functions call back to Mongo to get the full text of a story for geocoding. This makes it very difficult to tests. Consider splitting out the...
Mordecai returns both the raw place name extracted from the text, as well as the gazetteer entry it matches that place name to. Right now, the pipeline only has a...
Think about adding a pre-pipeline coding step that geocodes complete articles (rather than sentences) to the country. This would be useful for two things: 1. Associating actors that don't have...
To cut down on noise in the geolocation, we could consider only geolocating material conflict events (or in any case not geolocating statements and verbal cooperation). ¯_(ツ)_/¯