Social

Example of prototype of Social running on Python's socket, with jokes being displayed on terminal (optimally on pop-up messages in future!)

Inspiration

Find yourself struggling to make conversation? Want to improve your sense of humor? Well, then get SOCIAL! Making conversation is a difficult skill for everyone. In a world today that requires excellent conversational skills both in real-life and online, in the business world and in the dating scene, we figured we could make something that could help people grow their social skills!

What it does

Social is a program that automatically suggests jokes and topics to talk about based on your friend's-- or crush's ;) -- messages as you chat! And to let you know if you're doing well, we've included sentiment analysis, complete with indications of predicted emotions that your friend may be feeling!

How we built it

Using PyCharm, we built a chatroom with Python's own built-in Socket, and a GUI with Tkinter.

To create the joke and topic databases for automatic suggestions, we parsed and sorted databases scraped from reddit, stupidstuff, and wocka. Then, by parsing user messages into keywords, we used them to quickly suggest relevant jokes to say, or a new topic if a joke can't be found!

Using Python's NLTK library and scikit Learn to train Machine Learning models in order to classify textdata into emotions

Challenges we ran into

Making the machine learning models for sentiment analysis, as well as efficiently sizing down the databases to be able to search quickly

Accomplishments that we're proud of

Using Machine Learning!! And helping people impress their crushes ;)

What we learned

Socket is awful

What's next for Pocket Yoda

Smarter joke search engines, adding type of humor one wants to portray, catering suggestions to the type of person you're speaking to (parent, friend, girl/boyfriend)