AMR Parser for Bahasa Indonesia

AMR Parser from dependency parser features

Getting Started

Prerequisites

What things you need to install the software and how to install them

python3 
tensorflow

Installing

To install this app in your device, please do the following instructions

1. Install packages from the requirements.txt file

pip install -r requirements.txt

2. Create pretrained directory

3. Download this zipped_file that contains POS Tagger, NER Tagger, and Word2Vec models and put it in the pretrained folder. The tree structure should be like this:

pos_tagger/
ner_tagger/
word2vec/

4. Create saved_model directory

5. Download this pretrained model and encoders from this link and copy the contents of the file to the saved_model directory. The tree structure should be like this.

encoder/ 
---label_encoder.pickle.dat
---one_hot_encoder.pickle.dat
best_model_pretrained.pickle.dat

Running the program

Prediction

python3 amr_parser --predict --sentence <any sentence that you want> --output <filepath>

Example:

python3 amr_parser --predict --sentence "Aku makan buah di pagi ini" --output amr.txt

The output of the AMR should be included in the amr.txt file.

Training

COMING SOON

Authors

• Adylan Roaffa Ilmy
• Dr. Masayu Leylia Khodra, ST., MT. - Advisor