Project aim is to extract events from news articles. Now it's just bunch of heuristics on top of spacy grammar trees. We are working on it.

Bellow is short description of this project. Full project documentation absent, but we hope to make it someday.

We assume that event is a triple of (Subject, Action, Object). For example

todo add example

We have some crawlers written. So you can run

python3 spacy_event_extractor.py

to get about a couple of thousands articles from different websites. This script will also extract events just after it and put everything to database.

We have a simple web-UI to view and manipulate extracted events. You can start it just by running:

python3 run_web_server.py start

This command will start small server on port 9090. It can be accessed only from localhost.

We haven't made an install script yet, so it's all just a sequence of manual operations.

1. Dependencies

   You can install everything only by running

   pip install -r requirements.txt

   We are trying to keep it up to date.

   Note: we highly recommend to keep everything in separate virtualenv.

2. Database creation

   We are using postgres as our DBMS and depend on it. So you should have installed version of postgres. I have 9.3.14 installed on my system and everything seems to work

   psql postgres -f db/create.sql

   MacOS psql path:

   /Applications/Postgres.app/Contents/Versions/9.5/bin/psql postgres -f create.sql

3. Config creation

   Just copy default_config.cfg to config.cfg and set actual values to all fields.

   Note: DO NOT modify default_config.cfg as it can be changed in the future and you can lose your settings.