Twattle allows you to research the sentiment for brand or product based on a tweets from http://twitter.com. It is an academic project created by four Computer Science students in Web Science seminar at the University of Latvia. The project is used as a Twattle site at http://twattle.tk.

Twattle requires PHP 5.3 or newer, MySQL and MongoDB.

Clone the project

git clone git@github.com:ivoaz/twattle.git

Clone submodules

git submodule init
git submodule update

Copy the distributed config file and configure the databases

cp app/config/config.ini.dist app/config/config.ini

Add some objects to the database

from = new Date()
till = new Date()
till.setDay(till.getDay()+7)
db.objects.insert({
    "_id": "ipad",
    "name": "iPad",
    "keywords": ["ipad"],
    "topical_from": from,
    "topical_till": till
})

Run the collector service

php cli/collect.php

Processing tweets includes several operations - text normalization, object determination, language detection and sentiment analysis based on simple keywords method and Naive Bayesian method.

Before processing, make sure you have a lexicon in your mongodb database and statisical data for Naive Bayesian in your mysql database.

You can import default lexicon by running import script

php cli/import_lexicon.php

To start processing, run the processor script

php cli/process.php

Run the statistics script

php cli/stats.php

Configure your http server and open web/index.php in the browser.