Predicting a movie's genre(s) from its dialogue, using data from the Cornell Movie-Dialogs Corpus.
| File | Description |
|---|---|
main.ipynb |
Main notebook that loads the data, runs preprocessing, trains models, and reports results |
process.py |
Data preprocessing: builds character documents from dialogue, filters genres and stopwords, and converts text into unigram or average-embedding features |
train_eval.py |
Model training and evaluation, including a dev-set grid search over Naive Bayes and Logistic Regression configurations and final test-set scoring |
visualize.py |
Plotting utilities — genre distribution across movies, number of genres per movie, and a per-class error/confusion matrix |
paper.pdf |
Write-up describing the approach, experiments, and results |