Natural Language Processing for Finance

MSc Course | University of Twente

Course Overview

This course provides a comprehensive introduction to Natural Language Processing, covering the full spectrum from statistical foundations to advanced transformer architectures. Students learn to apply NLP techniques to financial text data including news, reports, social media, and regulatory documents.

Curriculum

The course is structured in eight chapters, each building on the previous:

Chapter 1 – The Statistical Era

Introduction to early methods such as n-grams, word frequencies, and probabilistic models like Hidden Markov Models (HMMs).

Chapter 2 – Neural Networks & Embeddings

Covers the transition to neural models and distributed word representations (e.g. Word2Vec, GloVe) that allow systems to capture similarity between words.

Chapter 3 – Sequential Models & Context

Explains how recurrent neural networks (RNNs), LSTMs, and GRUs can process sequences and learn dependencies in text.

Chapter 4 – The Transformer Architecture

Introduces the transformer model, self-attention mechanism, and encoder-decoder architecture, which are the basis of many modern NLP systems.

Chapter 5 – Text Classification

Shows how the above methods are used in real applications such as sentiment analysis, spam detection, and document labeling.

Chapter 6 – Generative Models

Focuses on how language models can generate text, including autoregressive models like GPT and masked language models like BERT.

Chapter 7 – Text Summarization

Explores how generative models are applied to produce shorter versions of longer texts, using both extractive and abstractive techniques.

Chapter 8 – Large Language Models (LLMs)

Covers the latest developments in models like GPT-3 and beyond, discussing scale, capabilities, limitations, and emerging business use cases.

Interactive Learning App

An interactive Streamlit app is available for exploring the course material:

NLP Evolution AppThe app provides a structured and interactive way to explore major developments in NLP. It is designed for students, researchers, and professionals interested in understanding how machines process and generate human language. Users can navigate freely or follow chapters in order.

Lecture Materials on Quantinar

All lecture materials are available on the Quantinar platform:

Learning Outcomes

• Apply NLP techniques to financial text data
• Build sentiment analysis models for market prediction
• Understand and implement transformer architectures
• Use pre-trained models (BERT, FinBERT, GPT) for financial document analysis
• Evaluate ethical implications of NLP systems in finance
• Deploy language models for practical financial applications

Target Audience

• Graduate and advanced undergraduate students in data science, computer science, business, or economics
• Researchers and educators interested in modern NLP systems
• Professionals working in analytics, finance, or policy

Prerequisites

• Machine Learning fundamentals
• Python programming
• Basic statistics