This repository holds the code for our group's main project. Our project is titled 'Enhancing BERT Sentiment Classification Syntactic Information.'
To install the dependencies, run pip install -r requirements.txt
from the main directory.
To train the model on the origional go-emotions dataset, run python main.py
. To train it on the "ekman" dataset, run python main.py --taxonomy ekman
, and to run it on the "group" dataset, run python main.py --taxonomy group
.
You can also choose whether it uses the origional go-emotions model or our modified go-emotions model via the --model_type
flag. By default, the modofied go-emotions model is used.
The syntax parser that we used was the Stanford Parser. It's not included in this repository because of file size constraints. However, the .jar
file can be downloaded from this link: Stanford Parser. Some of the parsed trees are included in the data
directory.
The paper and report for this experiment can be found in the repository with the file name report.pdf
.