DOC-AET: Improving Named Entity Linking through Anonymous Entity Mentions

Improving Named Entity Linking through Anonymous Entity Mentions.
Our code is based on the MulRel model Le and Titov 2018.

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 our manuscript.

Data

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

Evaluate

Download our pre-trained model from GoogleDrive

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

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 .