mulrel-nel: Multi-relational Named Entity Linking

A Python implementation of Multi-relatonal Named Entity Linking described in

[1] Phong Le and Ivan Titov (2018). Improving Entity Linking by Modeling Latent Relations between Mentions. ACL 2018.

Written and maintained by Phong Le (lephong.xyz [at] gmail.com )

Installation

• Requirements: Python 3.5 or 3.6, Pytorch 0.3, CUDA 7.5 or 8

Usage

The following instruction is for replicating the experiments reported in [1].

Data

Download data from here and unzip to the main folder (i.e. your-path/mulrel-nel).

Train

To train a 3-relation ment-norm model, from the main folder run

export PYTHONPATH=$PYTHONPATH:../
python -u -m nel.main --mode train --n_rels 3 --mulrel_type ment-norm --model_path model

Using a GTX 1080 Ti GPU it will take about 1 hour. The output is a model saved in two files: model.config and model.state_dict .

Evaluation

Execute

python -u -m nel.main --mode eval --model_path model