Natural Language Processing

About the course

The course balances theoretical foundations with practical examples in Python. No prior experience with libraries such as NLTK or scikit-learn is required. Experience with Python is very useful but not essential: people who use other languages and tools (Java, C++, C#, JavaScript, MATLAB, Excel or R) will also get a lot out of it.

What you will learn

Data representations for working effectively with text
Exploratory techniques to quickly gain insights from text data
Machine learning techniques to organise documents into categories
Evaluating the quality of your models
Ideas for advanced applications using natural language data

Syllabus

NLP foundations

The basic tools and techniques to get started with natural language processing.

NLP applications and the Python NLP ecosystem: NLTK, spaCy, Gensim, scikit-learn
Working with text: tokenisation, text pre-processing, regular expressions
Word frequencies and co-occurrences: stop-words and Zipf's law, mining topics of interest with co-occurrences
Text representation: n-grams, bag-of-words, word embeddings and document embeddings

Topic modelling

Understanding a document, or a collection of documents, with techniques that go beyond word frequencies.

A bird's-eye view of a document or a dataset
Navigating topics and sub-topics in a document or a dataset

Text classification

Classifying documents into a set of predefined categories.

Categorising documents
Topic classification
Sentiment analysis
Model evaluation: assessing classification quality
Model introspection: explaining classification results

Advanced applications

An outlook on more advanced NLP problems, so participants leave with ideas and techniques for specific applications, such as:

Named entity recognition: identifying named entities in text
Text summarisation: extracting the most useful sentences from one or more documents
Search engines: retrieving relevant documents from a custom text collection