This repository is used to build a multi-lingual system for identifying the factuality of events. We are currently working on a basic implementation for Dutch and English.
Running the module
The scripts to run the module are located at feature_extractor/
- On a single file:
cat inputfile | python rule_based_factuality.py > outputfile
- For a directory:
./run_rule_based_on_dir.sh inputdir/ outputdir/
Version 0.01 is released on May 9th 2016. It is a basic implementation and first creation of resources based on intuition. We expect to have many minor revisions in the near future. You will therefore find some version 0.xx in the document.
As soon as the resources for at least one language and the rule application system have stabilized, we will release version 1.0
This repository consists of:
docs/ for documentation data/ for development and training data resources/ for language specific resources (vocabularies) and models scripts/ for code extracting input features, applying rules/calling machine learning modules and producing output.
please check out: https://github.com/cltl/factuality_experimental_environment
For gold data, input data and evaluation scripts for factuality.