This is a set of data on learning using a flash-card techiques, generated by Mnemosyne users up to 2009/08/12. It was gathered from users who agreed (by setting proper option in Mnemosyne's configuration) to take part in research. This is not a deck of cards to be learnt; these files contain a history of grading, useful for statistical analysis and experimenting with new algorithms. The set is anonymized.
This set can be used for educational/research purposes. It was sent to me by Peter Bienstman, the main author of Mnemosyne. If you have any doubts about legal status of this dataset, please consult him.
The set was slightly modified; it originally consisted of about 52,000 files in single directory. Some operating systems/file managers might choke on such directory. Therefore all the files were divided into ~250 sets, each containing complete data from several randomly-chosen users, and packed using tar. If you are not interested in downloading all the data (900MB), you can use Azureus or other smart Bittorrent client to download only few prepacked files and treat that as a random sample.
The exact format of these files is defined in Mnemosyne's source code, which is opensource. It is not csv, you probably have to write some parsing routines by yourself. For more information please visit Mnemosyne website: http://mnemosyne-proj.org/
You are invited to discuss this dataset on Mnemosyne's public mailing list.