October 3rd was the release date of Super Smash Bros for 3DS - Nintendo’s newest game for its portable video-game console. Knowing this and wanting to practice some programming I decided to create a little pet project. "#SmashBros Twitter Analytics" is composed of two softwares: a service to capture and process Twitter data related to the game, and a website to show some interesting information about the tweets. Both were developed in Java platforms and had their code published on GitHub (https://github.com/danielgimenes/SmashBrosTwitterAnalytics).

The website is (for now) hosted on Amazon Web Services, and can be visited through the link: http://54.69.148.65/SmashBrosTwitterAnalytics/

smashbrostwitteranalytics_v1smash2

Libraries and Frameworks

SmashBrosTwitterStreamProcessor (service):

  • Java SE
  • Twitter4J
  • PostgreSQL JDBC

SmashBrosTwitterAnalytics (website):

  • Java EE 7
    • JSF
    • JPA
    • EJB
    • Filters
  • Glassfish
  • PrimeFaces
  • PostgreSQL JDBC

Rules for Capturing and Processing

The tweets captured were only the ones with the hashtag #SmashBros in it. This was configured on the Twitter Streaming query with Twitter4J.

Information Generated

The service processed every tweet and stored it on the relational database. This processing was designed to produce a rank of characters by references (of them on tweets) and a word count for every word in the tweets, excluding words from a list. That list included common english words (pronouns, etc) and words that wouldn't be interesting to analyse (like "Smash" for obvious reasons). The list can be found on the link: http://goo.gl/Y9fC2c

Also, the website displays the number of tweets captured since it was published, and the number of tweets grouped by hour and day on the last hours and days.