Skip to content

Latest commit

 

History

History

Sarcasm Detection

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
Dataset
Dataset
 
 
README.MD
README.MD
 
 
Sarcasm_detection.ipynb
Sarcasm_detection.ipynb
 
 

README.MD

SARCASM DETECTION MODEL USING MACHINE LEARNING(NLP)

Project Overview

The project is focused on building a sarcasm detection system using Machine Learning and Natural Language Processing (NLP) techniques. It uses the Sarcasm Headlines Dataset to classify news headlines as sarcastic or non-sarcastic. The system leverages text processing techniques and machine learning algorithms to identify and predict sarcasm in headlines.

Tools

  1. Programming Language: Python
  2. Libraries:
    • JSON: For handling and loading the dataset.
    • scikit-learn: Machine learning library for model building and evaluation.
    • TensorFlow: Used for deep learning operations (embedding and neural networks).
    • NLTK (Natural Language Toolkit): For text preprocessing and feature extraction.
    • Pandas & Numpy: Data handling and numerical operations.
    • Matplotlib & Seaborn: For data visualization and plotting results.

Processes

  1. Data Loading:

    • The dataset used is a JSON file (Sarcasm_Headlines_Dataset.json), which contains news headlines and a label indicating whether a headline is sarcastic (1) or not (0).
    • Initial loading of the dataset using Python’s json library.

  2. Data Preprocessing:

    • Tokenization and Vocabulary Building: Creating word sequences and defining a maximum length for each headline.
    • Padding and Truncation: Ensuring all sequences have the same length using padding.
    • Splitting the Dataset: The data is split into training and testing sets.

  3. Model Building:

    • The model uses various classifiers, including Random Forest and Deep Learning Models (e.g., a neural network in TensorFlow).
    • The Random Forest Classifier is implemented using sklearn to build a baseline for sarcasm detection.

  4. Evaluation:

    • Model performance is evaluated using accuracy metrics, confusion matrix, and F1-score.
    • Comparisons are made between different models to choose the most effective one.

Model Results (Random Forest Classifier)

  • Accuracy: The Random Forest Classifier achieved a decent accuracy on the test set.
  • Confusion Matrix: Showcases how many sarcastic headlines were correctly identified and the number of false positives/negatives.
  • F1-Score: The F1-score was used to measure the model’s precision and recall balance, demonstrating its efficiency in sarcasm detection.