CSCI 6040 Spring 2022
| Student | |
|---|---|
| Amy Wightman | dunphya20@students.ecu.edu |
| James Rosebaugh-Nordan | rosebaughnordanj20@students.ecu.edu |
| Connor Bullard | bullardc21@students.ecu.edu |
Authors' Repository
https://github.com/webis-de/emnlp21-same-sentiment
Our Implementation
https://github.com/ymadh/NLP/blob/main/Final%20Project/OurImplementation_v2.ipynb
The datasets referenced in this notebook use a portion of the kaggle yelp data files above which were created using head() in the shell.
To use the full dataset, run the kaggle test notebook which will download the files to the server you are using.
Dataset can also be downloaded directly from yelp using this link.
Failed Attempts & Previous Attempts
From the author's code with some minor changes to prepare for execution:
https://github.com/ymadh/NLP/blob/shared/D2_samesentiment_yelp_create_pairs.ipynb - This was successful
https://github.com/ymadh/NLP/blob/shared/D2_samesentiment_yelp_base.ipynb - Times out on the last cell. This requires you to run MLFlow in the backround (per the authors' README)
https://github.com/ymadh/NLP/tree/main/Final%20Project/Reproduction%20Attempt - Other files we were unsuccessful in modifying but attempted to use (some were Jupyter notebooks converted to Python scripts)
Optional
Login to kaggle and get API credentials. Update kaggle.json.
Run kaggle_test.ipynb notebook to get all the data files.