Group project work for CSCI 3832
Jake Henry Tuten (jatu5220@colorado.edu) Ashley Weaver (aswe8064@colorado.edu) flywilliunms (heri9386@colorado.edu)
Run "python 3832/main.py"
We used the training data given to us in the initial instructions of the project. The link to our training dataset is https://github.com/iabufarha/iSarcasmEval/blob/main/train/train.En.csv in the iSarcasmEval github folder. We only used the English version and not Arabic. We then used a mixture of code from homework 2 and outside sources to generate out model.
The goal of our project is using a preloaded dataset of tweets for training, our program will be able to detect what's sarcasm or not. We use the fundamentals we've learned in NLP (for example, Homework 2).
Out prediction.csv file is out output file located in 3832/prediction.csv. It has the tweets with the corresponding prediction. You can find the output of our F1 score once everything has run. Out best f1 score has been around 0.65. I include a picture of our model evaluation outputs in the github folder.