LSTM-powered language model to generate television scripts for "The Office".
By Rohit Awate, Shubham Bhagwat, Rohan Devasthale
This project aims to develop a natural language processing (NLP) model capable of generating a script for the popular TV series "The Office (US)". Utilizing dialogues from all nine seasons that aired on the show as a dataset, we experiment with various deep learning architectures, such as Bidirectional Long Short-Term Memory (LSTM) networks, deep neural networks. The resulting model will produce a script containing character names followed by their respective lines of dialogue, capturing the essence of the show's unique humor and style.
-
The
experiments/directory contains all the Jupyter notebooks that we worked on while working on this project. -
The
dataset/directory contains the CSV dataset that we obtained from Kaggle.
This is a class project executed as a part of the Computer Science course CS6120 "Natural Language Processing" at Northeastern University. Not affiliated with NBCUniversal and other copyright holders.