Sophomore year I took my first dive into the world of data work and natural language processing with Cornell's CS 4740: Introduction to Natural Language Processing taught by Professor Claire Cardie. While initially I may have been a little over my head I persisted and with the help of excellent teammates including Scott Cambo and Jennifer Doughty we created some pretty cool projects with some reasonably good results. Looking back, I have now realized there was some basic knowledge that I have after subsequent courses that I think could have really improved some of these
For this project I joined with a group of four students to create a bigram language modeling system. It was designed to be an initial understanding of language modeling, smoothing, and author detction. We faced several computational challenges as part of this project and our solutions are included in the writeup. To download and examine code click here
For this project I joined with a group of four students to create a deception detection system for hotel reviews. This is based on a paper by Professors Cardie and Hancock of Cornell University. The goal here was to design models to differentiate between deceptive and truthful hotel reviews (truthful reviews are reviews written by actual patrons of the hotel while deceptive reviews may have been written by paid services.) This project was an interesting case of feature selection in order to define what characteristics would show that a hotel review was deceptive. To download and examine code click here
For this project I joined with a group of four students to create a word sense disambiguation system. We did this both through bayesian analysis of the surrounding words and dictionary similarity.To download and examine code click here
For this project I joined with a group of four students to create Sentiment Analysis System. We were given a set of movie reviews by two authors and tasked with labeling the review on a scale from -2 to 2 on how much the reviewer liked the movie. We used several different methods for this an also created a letter grading system to classify the movies. To download and examine code click here
For this project I teamed with another student to explore the use of smart phone technology as a firefighter safety system. Our findings are here here