Punderstand.py

A program written to detect puns in a sentence using k-Nearest Neighbour and Naive Bayes Classifier.

This is made for the Fundamentals of AI final project on Pun detection, location, and analysis with different models.

Prerequisites

This program requires four dependencies, as listed below:

• datasets: Used to load and preprocess datasets.
• scikit-learn: Used for feature-extraction and machine learning models (kNN/NBC).
• nltk: NLP tools such as tokenization, text preprocessing.
• joblib: To persist trained models for reusability

Install the dependencies with

pip install -r requirements.txt

How to Run

Training the models

Running the main script will automatically train the two models and save them

python3 main.py

Making a prediction with the models

Running the script below with your model/input sentences to predict whether the sentence is pun-intended. When using kNN as model, you can also include n_neighbors flag to display a scattered plot of the input and its neighbors

python3 predict.py --model <model name> --input <text> --n_neighbors <optional number>