This repository contains the code for a stance detection project, which aims to identify the stance or sentiment expressed in text towards a target entity. The model classifies text as either supporting, opposing, or neutral towards a specific target, a task that can be useful for a variety of applications, including social media analysis, opinion mining, and information retrieval.
The goal of this project is to build a machine learning model that can predict the stance of a given text towards a target. The project uses a dataset of text samples and applies natural language processing (NLP) techniques to train the model.
- Stance Classification: Classifies text into three categories — Support, Against, or Neutral.
- Text Preprocessing: Includes tokenization, stopword removal, and text normalization.
- Model Implementation: Utilizes popular machine learning models and techniques for stance detection.
The dataset used for training and evaluation contains labeled text samples, each with a corresponding stance label. You can find the dataset in the /data directory or refer to the original source (if applicable).
- Clone the repository:
git clone https://github.com/ragini-kalvade/TweetStance
- Navigate into the project directory: cd stance-detection
- Install the required dependencies: pip install -r requirements.txt
Data Preparation: Preprocess the dataset using the provided scripts to clean and format the text.
Training the Model: Use the training script to train the model on the preprocessed data: python train_model.py
Making Predictions: Use the trained model to classify new text samples:
python predict.py --text "Your text here"
Example : Given a sample text:
text: I love this new policy and fully support it!
The model would classify it as: Support
Evaluation: The model's performance is evaluated using standard metrics such as accuracy, precision, recall, and F1 score. The evaluation script is available for assessing the results on the validation/test dataset. python evaluate.py
/stance-detection │ ├── data/ # Dataset files ├── models/ # Saved models ├── notebooks/ # Jupyter notebooks for exploration ├── src/ # Source code for preprocessing, training, and evaluation │ ├── preprocess.py # Data preprocessing script │ ├── train_model.py # Script to train the stance detection model │ ├── predict.py # Script to make predictions │ ├── evaluate.py # Script to evaluate the model └── requirements.txt # Python dependencies