/feedforward-RE

Feedforward NeuralNet for Relation Extraction

Primary LanguagePythonMIT LicenseMIT

feedforward-RE

Feedforward NeuralNet for Relation Extraction.

Data Preparation

  • Place data in CoType format obtained in data/source/$dataname.
  • Using KBP as the exmaple dataset, run ./brown_clustering.sh KBP to generate browncluster file (may take a while, don't repeat if nothing in data changed)

Feature Extraction

First, run the server using all jars in the CoreNLP home directory (in another terminal): java -mx4g -cp "*" edu.stanford.nlp.pipeline.StanfordCoreNLPServer -port 9000 -timeout 15000 Then: ./feature_extraction.sh KBP

Train FF-NN model

python3 FFNN/run.py KBP