Interactive Maps of Music, Movies, TVs and Books
Joint work with Emden Gansner, Stephen Kobourov, Stephen North & Mason Smith
Information visualization is essential in making sense out of large data sets. Often, high-dimensional data are visualized as a collection of points in 2-dimensional space through dimensionality reduction techniques. However these traditional methods often do not capture well the underlying structural information, clustering, and neighborhoods. GMap is an algorithm that turns these information into geographic maps, which are intuitive to most people. Using GMap algorithm and the OpenLayers library for map navigation give an interactive system for exploring the landscape of relations.
Map of Musics is based on a crawl of the last.fm website, from which we found which musician/group is similar to which other musicians/groups.
Map of Movies is based on data from the Netflix Prize, from which a measure of similarity between movies are calculated (similarity matrix kindly provided by Yehuda Koren).
Map of TVs is based on aggregated and
anonymized TV viewing data.
Map of Books is based on a crawl of the Amazon website, from which we found which book is similar to which other books, via "customer who bought x also bought y" information available from the site.
Map of the Internet is based on data from xmarks.com , from which we found which website is similar to which other sites. (Joint work of Daisuke Mashima, Yifan Hu and Stephen Kobourov)