Irony-and-Sarcasm-Detection

URL: https://github.com/ronanmmurphy/Irony-and-Sarcasm-Detection

Description: Trained a Logistic Regression and Recurrent Neural Network (with LSTM layer) models to perform classification of Ironic and Sarcastic Tweets using Natural Language Processing. Pre-processing of the text data and word vectorisation was required before training these models using TensorFlow. Evaluated the models’ performance based on Precision, Recall, and F-Scores. The sequential model produced accuracy over 90% for training, validation, and test datasets.

Methods

• Model Types: Logistic Regression and Recurrent Neural Network (RNN) with an LSTM (Long Short-Term Memory) layer.
• Text Preprocessing: Includes data cleaning and word vectorization to convert text into numerical format suitable for machine learning models.
• Training Framework: TensorFlow was used for training both models.
• Evaluation Metrics: Performance was evaluated using Precision, Recall, and F-Scores.

Results

• Performance: The sequential model achieved an accuracy of over 90% on training, validation, and test datasets.
• Evaluation Metrics: Detailed metrics include Precision, Recall, and F-Scores (specific values not mentioned, but the model is reported to perform well).

Dataset

• Dataset Type: Dataset consists of tweets classified as ironic or sarcastic.
• Data Preprocessing: Required steps include text pre-processing and word vectorization to convert the text data into a usable format for the models.