MoodMap

Plotting sentiment analysed tweets to show the mood of the country

MoodMap is an online application which correlates data from Twitter with data from the government. Tweets are put through sentiment analysis (to assess the overall mood) and then plotted on a map according to the location from which they were tweeted from. Goverment data for deprivation is then overlayed on top of this.

Authors

• Priyesh Patel
• Daniel Saul

Screenshots

Setting up

Dependencies

All dependencies are listed in requirements.txt. Inside your VirtualEnv execute:

pip install -r requirements.txt

Installation

It is recommended to run the MoodMap daemon with Supervisor. To do this setup Foreman by copying env.example to .env and filling out the details. Foreman can then be used to export a Supervisor configuration file:

foreman export supervisord ./

This resulting file will need to be edited so that /bin/sh venv_<process>.sh is run as the command in each case and then added to the system's Supervisor directory (often /etc/supervisor/conf.d/).

Supervisor can then be reloaded with sudo supervisorctl reload.

The web root should be set to ./web.

CouchDB design docs are provided in ./couchdb.

ANEW Dataset

MoodMap utilises the ANEW (Affective Norms for English Words) dataset produced by the University of Florida. The data is also available from other sources: http://www.manifestdensity.net/2010/06/16/anew/.

The data should be imported into CouchDB and each doc should be formatted as such:

{
    "_id": XXX,
    "_rev": XXX,
    "word": "ball",
    "value": 6.00
}