As a winter project, I'm attempting to use a Recursive Neural Network to generate an episode script from the long-running animated show The Simpsons. At the time of this writing, there have been 627 episodes of The Simpsons, which provides a substantial corpus of data to be used in training the RNN.
The project will be using the TensorFlow library to create and train the RNN, the Python package BeautifulSoup to perform web-scraping for retrieving the scripts, and, as a stretch goal, intends to use Go to efficiently tokenize the scripts into feature vectors to be used by TensorFlow.
Python 3.6.3
Tensorflow 1.4.0
beautifulsoup4 - 4.6.0
12/23/17 - Complete work on tokenizing scripts, which maps the ten-thousand most common words to an integer, or -1 if less common.
12/22/17 - Begin work on creating converting scripts to machine friendly representations
12/16/17 - Initial Commit