David Kramer's high-entropy blog

Portfolio: STAR

STAR is an Aptima project that breaks news articles from multiple sources into semantic topic profiles (a group of topics and their probability), allowing analysts to look for patterns of those topics, with an emphasis on identifying leading indicator topics for other topics. The point is to be able to predict when a particular topic will be prevalent in the next few days by looking at those topic patterns. A collaboration mechanism lets analysts share their thoughts on the program’s predictions, and see how the predictions would change if the analysts predictions were factored in.

STAR is based on several tiers separate components. The heart of the application is Java/J2EE with JSP pages for the view layer.  Adobe Flex is used for the many graphs produced by the system (created by a co-worker).  However, most of the application resides on one page which relies heavily on JavaScript, AJAX, DOM manipulation, and event handling to visualize and hide content as the user enters inquiries.


Site info