Information Forensics

This grant led to several accomplishments. One of the more important results involved web search. We developed a technique of ranking web pages that was unaffected by changing links on a web page. We proved that you could not increase your rank by creating short cycles, nor could you hurt the rank of another web page by more than you had increased the pages rank. We also developed techniques for determining which pages contributed most to a given pages page rank. In the early years of the World Wide Web, individuals put links on their pages for navigational purposes. However, as Google and other search engines become very efficient at locating web pages, individuals ceased putting links on their pages for navigational purposes. We developed a theory to show what impact this would ultimately have on the webs structure. Other areas we contributed to are as follows the development of multi-task learning, techniques for finding communities in graphs, and tracking the flow of ideas in scientific literature. Some initial progress also has been made in determining the connections between people based on when they acquired an item.