Text Analysis

This repository contains a collection of notebooks and resources for various NLP tasks using different architectures and frameworks. Each notebook focuses on a specific task and demonstrates the implementation using appropriate datasets and models.

Notebooks

1. Bidirectional-Stacked-LSTM.ipynb

• Task: Text Classification using RNN-LSTM
• Direction: Bi-directional
• Layer Type: Stacked
• Framework: TensorFlow with Keras
• Dataset: IMdB movie review

This notebook demonstrates text classification using a bidirectional stacked LSTM architecture. It uses the IMdB movie review dataset for training and evaluation.

2. LanguageModelling.ipynb

• Task: Language Modeling (LM)
• Architecture: RNN / LSTM
• Framework: TensorFlow with Keras

This notebook focuses on language modeling using RNN / LSTM architectures. It demonstrates the training process and evaluates the model's performance.

3. MaskedLanguageModelling.ipynb

• Task: Masked Language Modeling (MLM)
• Architecture: DistilRoBERTa
• Dataset: ELI5 dataset
• Framework: PyTorch using HuggingFace

This notebook covers masked language modeling using the DistilRoBERTa architecture and the ELI5 dataset. It utilizes the HuggingFace library for model implementation and training.

4. NamedEntityRecognition.ipynb

• Task: Sequence Labeling - Named Entity Recognition (NER)
• Architecture: Transformer
• Dataset: CoNLL 2003
• Framework: Keras

This notebook demonstrates named entity recognition using transformer architectures. It uses the CoNLL 2003 dataset and Keras for model implementation.

5. QuestionAnsweringSystem.ipynb

• Task: Question Answering (QA)
• Architecture: BERT - Transformers
• Dataset: SQuAD
• Framework: TensorFlow with Keras / PyTorch (HuggingFace)

This notebook focuses on building a question-answering system using BERT and transformer models. The SQuAD dataset is used for training and evaluation, with implementations available in both TensorFlow with Keras and PyTorch using HuggingFace.

6. Single-layer-LSTM.ipynb

• Task: Text Classification using RNN-LSTM
• Direction: Sequential
• Layer Type: Single
• Framework: TensorFlow with Keras
• Dataset: IMdB movie review

This notebook provides a simpler text classification example using a single-layer RNN-LSTM architecture. It also uses the IMdB movie review dataset.

7. TextClassificationUsingTransformers.ipynb

• Task: Sequence Classification
• Architecture: BERT - Transformers
• Dataset: IMdB
• Framework: TensorFlow with Keras / PyTorch (HuggingFace)

This notebook covers text classification using BERT transformer models on the IMdB dataset. It includes implementations in TensorFlow with Keras and PyTorch using HuggingFace.

8. XOR-Scikit-Learn.ipynb

This notebook demonstrates the XOR problem using Scikit-Learn. It provides a basic example of solving a classic machine learning problem.

Additional Files

wonderland.txt

This text file is used for language modeling tasks. It contains text data for training language models.

Getting Started

1. Clone the repository:

   git clone https://github.com/selcia25/text-analysis.git
   cd text-analysis

2. Run the Notebooks: Open the notebooks using Jupyter or any other compatible environment and run the cells to train and evaluate the models.

Requirements

• Python
• TensorFlow
• Keras
• PyTorch
• HuggingFace Transformers
• Scikit-Learn
• Jupyter Notebook

License

This project is licensed under the MIT License.

Acknowledgements

Special thanks to the contributors and the open-source community for providing the datasets and frameworks used in these notebooks.