Part I: Fundamentals
- Preprocessing: tokenization, simplification techniques, tagging,and simple rules-based approaches.
- Basic Vectorization: turning text into numbers and measuring similarity between documents.
- Modelling Overview: types of machine learning, algorithms → vs. models, evaluation, ...
- First Steps into Classification: ... classifying text using Naive Bayes; evaluation with precision and recall.
- Topic modelling: Automatically finding topics in documents using Latent Dirichlet Allocation.
Part II: Deep Learning for NLP
- Neural Networks: what they are, how they work, and details around training and evaluation.
- Word Vectors: capturing word meaning; the concept of embeddings.
- Recurrent Neural Networks: capturing sequence information and generating language.
- Seq2seq and Attention: training a neural network to transform one sequence into another.
- Transformers: the dominant mainstream architecture today; pretraining and transfer learning.