Posted with permission from instructor.
Authors/Creators: Matthew Liechty and N. Lee
Based on the paper Humor Detection in Yelp reviews by Luke de Oliveira and Alfredo Lainez Rodrigo of Stanford University.
The write up for this project and its results can be viewed here.
The code written to generate the results for this project (besides some of the data pre-processing code) is available in the Jupyter Notebook project_code.ipynb.
In this project we used a wide variety of neural network approaches to learning humor from Yelp data. Approaches include feedforward neural networks, bidirectional LSTMS, bidirectional GRUs, and convolutional neural networks (CNNs).