Context Is(n't) King: Named Entity Recognition Based Solely on Surrounding Words

Second Year Project (Introduction to Natural Language Processing and Deep Learning)

Authors

• Christian Hetling, chrhe@itu.dk
• Krzysztof Parocki, krpa@itu.dk
• Malthe Have Musaeus, mhmu@itu.dk

Getting started

1. Start by cloning the repo: git clone https://github.com/Hetling/NLP-second-year-project.git
2. Download the contextualized word embedding pickle files here: https://drive.google.com/drive/folders/1SinJt4EaPbn2el-Yjhj_KN7KkaCW7LY2
3. Create a new models folder in the root directory
4. Place the downloaded data folder inside of the newly created models folder.

Usage

Now you are ready to train, validate, and test the models. The main.py file acts as a simple CLI to interact with the models. The usage of which is described below:

To train all models and save them to disk

  python main.py --train

To train only approach 1 and 2 without saving them

  python main.py --train --approach-1 --approach-2 --save False

To validate all models from disk. Remember to train them first

  python main.py --validate

To validate only approach 1 and 2

  python main.py --validate --approach-1 --approach-2

To test all models from disk. Again remember to train them first

  python main.py --test

To test only approach 1 and 2

  python main.py --test --approach-1 --approach-2

To train the baseline model, run baseline.py without any arguments. To reproduce visualizations, see visualize_net.ipynb