Skip to content

Latest commit

 

History

History
39 lines (32 loc) · 2.57 KB

README.md

File metadata and controls

39 lines (32 loc) · 2.57 KB
Raw

Revisiting Joint Modeling of Cross-document Entity and Event Coreference Resolution

Introduction

This code for EECDCR was originally used in the paper:

"Revisiting Joint Modeling of Cross-document Entity and Event Coreference Resolution"
Shany Barhom, Vered Shwartz, Alon Eirew, Michael Bugert, Nils Reimers and Ido Dagan. ACL 2019. (https://www.aclweb.org/anthology/P19-1409/)

A neural model implemented in PyTorch for resolving cross-document entity and event coreference. The model was trained and evaluated on the ECB+ corpus.

The code belongs to Shany Barhom (shanyb21@gmail.com) and was used inside the Xcoref as one of the baselines. Original github project is https://github.com/shanybar/event_entity_coref_ecb_plus. Please contact Shany for questions about the original model and code.

Setup for the CDCR pipeline

  • PyTorch
    • PyTorch 0.4.0 with CUDA 9.0 on Linux can be installed using the command: pip install https://download.pytorch.org/whl/cu90/torch-0.4.0-cp36-cp36m-linux_x86_64.whl
    • torch 1.2.0 and torchvision 0.4.0 for different OS: https://pytorch.org/get-started/previous-versions/#v120 pip install torch==1.2.0+cu92 torchvision==0.4.0+cu92 -f https://download.pytorch.org/whl/torch_stable.html --user (if you experience troubles, try without "user" option and then with it) If torchvision does not show up in a list of installed packages, run pip install torchvision and then rerun the installation with the version above.
  • AllenNLP pip install allennlp==0.9.0

Download models and libraries

Download the following archive, exract the files, and place the models cd_entity_best_model and cd_event_best_model into ./resources/eecdcr_models.

The other model files will be downloaded directly from the code or can be downloaded manually:

  • ELMO
  1. weights
  2. options
  3. Place both files into ./resources/word_vector_models/ELMO_Original_55B
  • BERT SRL
  1. model
  2. Place the model into ./resources/word_vector_models/BERT_SRL