buaadk / KJETE

A KNOWLEDGE/DATA ENHANCED METHOD FOR JOINT EVENT AND TEMPORAL RELATION EXTRACTION implementation code
2 stars 0 forks source link

Date: April 15nd, 2022

Title: Code for Paper: [A KNOWLEDGE/DATA ENHANCED METHOD FOR JOINT EVENT AND TEMPORAL RELATION EXTRACTION]

The TB-Dense and MATRES raw data files are saved in data fold. Download any other necessary files into other/ folder.

  1. Before the experiment, modify the configuration file to your own path

  2. Data processinng. run python addcommonsense_featurized_dataall*.py first,the folder all_joint/ are created

  3. Featurize data. run python contextaggregator*.py. The folder all_context/ are created: .

the step 2 all_context contains the final files used in the model.

  1. Local Model: run python joint_model_addcommonseStandModelroberta*.py

For convenience and ease of understanding,You can run it separately on both datasets.