- Netflix has created a contest to see if anyone can improve their movie recommendation algorithm. To this end, they have enabled contestants (free registration required) to download user rating data for 17700 movies, comprising approximately 31GB in all. The bonus to using this data set is that if you improve their algorithm by 10%, they'll give you $1,000,000.
- The United Nations has unveiled recently a unified data site with very extensive data from its member nations over a broad array of topics. The data is available for download in variety of formats.
- Dataset, from the libsvm website. Constains several sets of data from several sources and various numbers of features.
