Clustering & Topic Modelling
Topic modelling aims to discover content-related patterns, or topics, in a collection of texts. It has commonalities with clustering techniques that group documents in an unsupervised way, and which we also cover here.
Lecture Slides
to be added
Lab
In this lab, you will look at k-means clustering of product reviews, and train a topic model on a dataset of US political speeches.
Reading Material
- Blei (2012), a review article on topic modelling
- Chapters 16 & 17 of Manning, Raghavan and Schütze (2008) on clustering