We were originally going to perform sentiment analysis and topic modelling to determine the best tips for recipes. However, this proved too much for our time frame, so we are focusing on the sentiment analysis. We're building multiple sentiment analyzers and contrasting their results against our hand-coded data for one recipe: the basic NLTK Vader sentiment analyzer, and three RNNs using the imdb and food.com training data sets and an additional word embedding. We'll also hope to share some preliminary results of n-gram topic modelling.