Course Outline

🔹 Day 1: Introduction to NLP and Text Preprocessing

Session 1: What is NLP and Why It Matters

• Overview of NLP and its applications
• NLP vs. traditional text processing
• Challenges in understanding human language (ambiguity, context, grammar)

Session 2: Text Cleaning and Tokenization

• Removing noise: punctuation, stop words, case normalization
• Tokenization techniques (word, sentence)
• Regular expressions for pattern matching

Session 3: Linguistic Features and Text Normalization

• Stemming vs. Lemmatization
• POS tagging and syntactic parsing
• spaCy and NLTK for linguistic analysis

Lab Activities:

• Clean and tokenize real-world text (news articles, tweets)
• POS tag and lemmatize sentences using spaCy
• Create a preprocessing pipeline from raw text to normalized tokens

🔹 Day 2: Feature Extraction and Core NLP Tasks

Session 1: Text Representation Techniques

• Bag-of-Words and TF-IDF
• N-grams and context windows
• Vectorization with scikit-learn

Session 2: Text Classification

• Sentiment analysis using logistic regression or Naive Bayes
• Multi-class classification (e.g., spam detection, topic labeling)
• Model training, testing, and evaluation

Session 3: Named Entity Recognition and Text Similarity

• Rule-based and statistical NER
• Jaccard, cosine similarity, and sentence similarity measures
• Text clustering and grouping

Lab Activities:

• Convert a dataset into TF-IDF features
• Train and evaluate a classifier for movie reviews or tweets
• Use spaCy to extract named entities from legal or financial text

🔹 Day 3: Sequence Models, Transformers, and Project Showcase

Session 1: Introduction to Sequence Models

• Word embeddings (Word2Vec, GloVe)
• Recurrent Neural Networks (brief overview)
• Using pre-trained embeddings for NLP tasks

Session 2: Transformers and Pretrained Language Models

• What are Transformers and why they matter in NLP
• Using Hugging Face Transformers (BERT, RoBERTa, DistilBERT)
• Fine-tuning vs. zero-shot classification

Session 3: Capstone Project + Final Reflections

• End-to-end NLP project (e.g., customer feedback analysis or support ticket routing)
• Presentation and peer feedback
• Future of NLP: Multilingual models, GPT, ethical NLP design

Lab Activities:

• Load and apply a BERT model using Hugging Face for classification
• Compare traditional ML vs. transformer-based performance
• Build a working NLP pipeline from input to output