Project Update 2

[shantanu-555]

Project Update 2

More Visualizations

We added more visualizations by making a line-plot for the number of dialogues they spoke broken down by season to show the trends in a character's relative relevance across the show.

A Dialogue Generator based on Markov Chains

A Web-App that hosts our project for users to interact with the data

We created an interactive web app that hosts our project. It allows the users to interact with the data and the models deployed. Specific pages and nuances of the app are described in this section below. The app is created using a python framework called streamlit (https://streamlit.io/). To host the app yourself, clone the repository to your local system and run the following command:

streamlit run .\app\Home.py

Note that for the app to run as expected, you would have to install the following dependencies:

pip install streamlit
pip install markovify
pip install plotly
pip install wordcloud

Currently, the app contains a home page, a wordcloud generator, and a random dialogue generator.

Home page

The home page shows a the preliminary analysis of the data mentioned in Initial analysis of the data and visualizations.

Dialogue Generator page

This page hosts the dialogue generator described in A Dialogue Generator based on Markov Chains

It allows the user to select a character from the drop-down menu and runs the dialogue generator model. The app caches the data for a particular character when run for the first time so it doesn't have to train the model again when run for the same character.

Word-cloud Generator page

This page hosts the code for generating a word-cloud of the most spoken words by a character as described in Initial analysis of the data and visualizations.

Users can select a character from the drop-down menu and the app caches the model so it can be loaded faster when called again.

Sentiment Analyzer page