Chatterbot implementation using Transfer Learning and LSTM Encoder-Decoder Architecture

Abstract

The goal of this project is to develop a chatbot using deep learning models. Chatterbot is an existing research area whose main goal is to appear as human as possible and most of the current models(which use RNN and related sequential learning models) are unable to achieve this task to relate over long dependencies. Adding onto that, NLP tasks require a lot of data which can be hard to collect for smaller projects/tasks which inspired to try out sequence to sequence learning model using LSTM. For that we have used a movie dialog corpus of 220,579 conversation exchanges among which about 50,000 conversational exchanges are only used as a training corpus to our model since training on more conversation exchanges requires high computation power than we have.