Description

When I was on the HackYale team, we taught courses on web development, C++, graphic design, and android development. As an instructor, I decided to develop an accessible, three-week course on Natural Language Processing in Python.

The Problem

Yale’s Computer Science department focuses on fundamental topics in C and LISP. When HackYale was founded in 2011, there was a desire on campus for courses on additional topics outside the scope of the major.

Specifically, there was a strong need for courses in web development and high-level programming languages. In fact, our Introduction to Web Development course regularly had 200+ applicants for ~30 spots.

Planning

I wanted the course to be accessible to students without CS backgrounds. The hardest part was to describe NLP concepts in a straightforward way. I used the following resources when planning out my curriculum:

• Speech and Language Processing
• Coursera’s Natural Language Processing MOOC
• Python’s Natural Language Toolkit docs
• Personal notes from LING227: Language and Computation I

Outcomes

I taught the class to a group of CS students, an MBA from the Yale School of Management, and a PhD student in Astrophysics who wanted to retool as a Data Scientist.

Over the course of three weeks, we covered corpus manipulation and analysis, n-gram language models, naive bayes text classification, and concluded by categorizing a dataset of news articles using a model we built with the Reuters Text Classification Corpus.