Natural Language Processing (NLP) Projects

This repository is a collection of my Natural Language Processing (NLP) projects and research experiments.
It covers work ranging from classical methods to modern transformer-based models and LLM (Large Language Model) research.

---

Repository Structure

📦 NLP ┣ 📂 Sentiment_Analysis ┃ ┣ Sentiment_Analysis_using_MachineLearning.ipynb ┃ ┣ Sentiment_Analysis_using_VADER_and_RoBERTa.ipynb ┃ ┣ Fine_Tuning_BERT_for_Sentiment_Classification.ipynb ┃ ┗ Explainable_Sentiment_Classification_using_LIME_and_SHAP_for_BERT_based_Model.ipynb ┣ 📂 Text_Summarization ┃ ┣ Text_Summarization.ipynb ┃ ┣ text_summarization_using_seq2seq_mpdel.ipynb ┃ ┣ Text_Summarization_using_T5.ipynb ┃ ┗ Abstractive_Text_Summarization_using_T5_BART_on_XSum_with_ROUGE_Evaluation.ipynb ┣ 📂 Question_Answering ┃ ┣ Question_Answering_using_Transformer.ipynb ┃ ┗ Knowledge_Graph_Backed_QA_System.ipynb ┣ 📂 Machine_Translation ┃ ┗ Machine_translation_using_transformer.ipynb ┣ 📂 Embeddings ┃ ┗ Word2vec.ipynb ┣ 📂 LLM_Research ┃ ┣ Probing_LLM_Vulnerabilities_A_Red_Teaming_Mini_Case_Study.ipynb ┃ ┗ LoRA_Based_Pruning_Evaluation.ipynb ┗ 📂 Basics ┣ Basic_NLP_(Classical_Approach).ipynb ┗ Natural_Language_Processing.ipynb

---

Key Highlights

Sentiment Analysis

• Classical ML approaches (Naive Bayes, SVM, Logistic Regression).
• Lexicon-based approach using VADER.
• Transformer-based RoBERTa for sentiment classification.
• Explainable AI: LIME & SHAP applied to BERT-based models.

Text Summarization

• Classical extractive summarization.
• Seq2Seq with attention.
• Abstractive summarization using T5 and BART.
• Evaluation using ROUGE metrics.

Question Answering

• Transformer-based QA systems.
• Knowledge Graph-enhanced QA for structured reasoning.

Machine Translation

• Transformer-based English ↔ Other language translation.

Embeddings

• Word2Vec representation and vector space exploration.

LLM Research

• Red Teaming Case Study: Probing vulnerabilities in LLMs.
• LoRA Pruning & Evaluation for parameter-efficient fine-tuning.
• Bias detection & mitigation experiments.

Basics

• Preprocessing pipelines (tokenization, stopwords, POS tagging, TF-IDF).
• Classical NLP workflows.

---

Tech Stack

• Languages: Python
• Frameworks & Libraries: PyTorch, TensorFlow/Keras, Hugging Face Transformers, Scikit-learn, NLTK, SpaCy, Gensim
• Tools: Jupyter, Colab, Matplotlib/Seaborn for visualization

---

How to Use

Clone the repository:

git clone https://github.com/YourUsername/NLP.git
cd NLP
Launch Jupyter Notebook:

bash
Copy code
jupyter notebook
Or open directly in Google Colab (recommended for GPU):

markdown
Copy code
[![Open in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/YourUsername/NLP)