Innovations in Twitter Inference and Analysis Techniques: Breakthroughs from Final-Year Undergraduate Computer Science Project

This technical report highlights several advancements in Twitter inference, analysis, and visualization techniques, produced as part of the fit3036 Computer Science Project unit (Semester 2, 2012)1; under the supervision of the corresponding author. Firstly, the successful deployment of a Web-based Twitter analysis and visualization interface on the Google Application Engine platform. The second is a frequency-based approach to keyword analysis, to discover similar terms related to a particular set of tweets, and its potential for targeted advertising. The third contribution consists of a significant improvement to the gender inference algorithm to determine genders of Twitter users based on their Twitter handles. The final contribution is an approach to detecting localized tweets and to compare the emergent features of tweets and authors across different locales. The findings from this report, culminating from a combined effort of 48 weeks of research from FIT3036, have improved the body of knowledge on approaches in Twitter analysis, and pave the way for future developments in related areas.