Traveling through the Internet, users generate a wealth of data that can be used to analyze habits, suggest new paths of discovery and support behavior.

Today, several popular web browsers have taken advantage of part of this data stream by providing thumbnail views of websites individuals have accessed recently. Simple algorithms based on the frequency of access to a particular URL gives way to the suggestions made available to users. It has become a very useful, very powerful tool. Bubble Stacks seek to improve upon this concept by opening the floodgates to utilize not just a simple stream of time/frequency data, but organize behavior based on a much wider realm of available information.

Among other potential data sources, Bubble Stacks are supported by the following elements associated with a given website (URL):

• Time of access
• Frequency of access
• Onscreen text
• Page titles
• Meta tags

In addition to time based analysis, we have proposed the notion of tapping into the information already available on each page itself. Onscreen text doesn’t get us very far by itself because to automatically analyze the information available, a system would need to understand the context of use. Understanding context, and thus concept is no small feat of artificial intelligence, but incredible advancements in this area have already been made.

One possibility of determining context is Princeton University’s WordNet. A number of incredible projects have been created tapping into its potential for recognizing connections between words and concepts. Other systems of artificial system intelligence exist. WordNet is simply a highly accessible, widely known system.