WordsWordsWords

Screenshot of our app in action

Inspiration

The autocomplete function on our phones keybord was our main inspiration for the app idea. However, we knew that we wanted to use an API hosted on Google cloud for the back end, and we really wanted to try using new technologies, so this was a simple enough project where we could learn cloud computing for the first time.

What it does

It generates (somewhat meaningless) sentences that sound Shakespearean. Mostly for comic effect

How we built it

First, we used a webscraper to find all the text from all of Shakespeare's work. Then we built a Markov Chain model to find a list of suggested next words given a current word. Then we stores all of this data on a mongodb cluster. Then we built a Flask API to access data from this database. Then we deployed the flask API on Google cloud. Then we built a React frontend to interact with the API using http requests.

Challenges we ran into

We did not know how to deploy the API on Google cloud for a long time. We tried deploying it on an app engine first, but that didn't work because it wasn't able to access data from the database. Then we moved to a compute engine and that solution worked. Also react is a very disgusting framework to work with. We also struggled for a while with getting our API to run on the compute engine without having the ssh she'll open.

Accomplishments that we're proud of

Deploying our first ever API on Google Cloud (and all the learning that required). Using mongodb for the first time. Successfully making a full stack app (flask, mongo, react).

What we learned

All the work involved in making a full stack app.

What's next for WordsWordsWords

More authors!!!! And better UI design for mobile. Handling mixed content errors.

HOW TO TRY IT OUT

On chrome: click the link below (amanj120.github.io), and on the top right corner, click on the little shield and click "allow unsfae scripts". The reason for that is because the website is a https website, but we are using http requests.

Built With

Submitted to

HackGSU - Spring 2019

Created by

I created the webscraper and made the initial API. I also helped debug the react frontend code a little. And I helped set up the mongodb cluster, and accessing that. I also messed around with Google cloud for six hours before passing it off to Justin. (I got it working on an app engine but it was really slow)

Aman Jain
I worked on the frontend, creating the user interface in React. I worked on retrieving data through the backend's endpoints and displaying them to the user, as well as handling user interaction logic.

Jeffrey Luo
I worked on the backend, generating the Markov chain and writing the CL utilities that wrote the resultant data to files or databases. I wrote the routes to interface with the Mongo database running on Cloud. I also set up the project to run in a VM on Google Cloud Compute Engine, and wrote supervisor utilities to keep it running.

Justin Prindle

Updates

Aman Jain started this project — Mar 31, 2019 05:36 AM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.