07 Feb 2013

Twitter Faces

Does social networks have emotions? For this project, I would like to experiment and try to put a face and extract expressions from Tweets around the world. The idea is to get a decent amount of recent tweets around the center of a big city, for example New York, do a basic sentiment analysis and extract an average mood of the people posting. The application will run in real time and I will query for new tweets every 5 or 10 minutes, this way I could create a lively face, that constantly changes mood and expressions. I think it will be really interesting to see the overall mood of people in a certain city, are they happy when there is a big event coming up, or are they sad, if something bad has happend. For the moment, I will use a really basic sentiment analysis, using only dictionaries with positive and negative words, but along the way, I could switch to a more complex method, if I find something suitable for real time processing.

Here is just a quick sketch I did in Illustrator, just to show how the application might look. The idea is that the face and expression changes will be animated, so we could get that lively feel. Also, I could list the most mentioned positive or negative words.

Twitter Faces

I will probably implement the app in Openframeworks, but I am also thinking about doing an online version, probably using Processing.js

Any comments are welcomed!

2 thoughts on “Project 2: Twitter Faces – Concept

  1. Golan

    Hi Kamen!
    This looks like a good idea, I like your sketch :). I think it would be more easily shared using Processing, though you’d still have to deal with the issues of connecting to a server script that could do the Twitter API handling for you. Unfortunately Twitter imposes tight limits on how much data you can pull — watch out, don’t get locked out by making too many requests. If you like, I have some example code in Processing for communicating with Twitter; see this page (at the top, and then about halfway down) from my class last semester: http://cmuems.com/2012/a/lectures/lecture-11/
    There are tools you can use for sentiment analysis, so you don’t have to create something from scratch; check out http://www.quora.com/Where-can-I-find-an-online-API-for-sentiment-analysis
    I’m not sure what the advantage of using OpenFrameworks would be, in this case; I think Java and JS offer more diverse tools for connecting to the network and doing text processing.

  2. ~Kamen Post author

    Hey Golan,
    Thanks for the input. Actually, I already have a working example, which I will put online and do a separate post about. I ended using Processing.JS for visualisation and server side Node.js for handling the Twitter API requests and the sentiment analysis.

Comments are closed.