Word Vectors and Corpus Text Mining with Python

How could we examine language and culture through computation? This workshop is an in-depth and hands-on overview on computational text mining and word vectors. It will cover from basic text mining foundations of Natural Language Processing and word vectorization, to the most recent advances at the intersection of computational linguistics and humanist applications. We will guide attendees through the theoretical and conceptual background necessary for understanding commonly used tools in DH such as Word2Vec and LDA Topic Modeling. The group will implement techniques and learn to critically analyze and evaluate results. There will be time to reflect on catering these applications and tools to participants’ specializations. Some programming experience is a prerequisite (not necessarily Python) and we will spend a session reviewing the Python necessary for the workshop and the Jupyter environment. Participants are encouraged to bring their own corpora of investigation for the hands-on activities.

Week 1:

This week will set the foundations for participants to be able to work with large-scale humanist text data, starting with data-wrangling and finishing with concepts and tools in machine learning. By the end of this week, participants will be comfortable with simple techniques such as word counts and building simple n-gram models, as well as discussing basic machine learning concepts for humanist applications.

Week 2: 

This week will focus on a popularized method in recent DH scholarship: word vectors. We will cover the concepts and assumptions behind traditional word vectors, Word2Vec and other neural net-based vectorization techniques, as well as their implications for the study of the humanities, e.g., is all bias bad? What can we learn from it? We’ll also spend time reviewing works in the latest DH literature and critically analyzing their methods. The focus will be on participants’ analytical engagement of these methods to their own data and critical interpretation of these applications on humanist inquiries.