This archive contains a large number of kidney exchange instances, each modelled as a JSON file. The files are named Generated_notes_20140709__-.json (e.g., Generated_notes_20140709_50_0.05_5.json). Size denotes the number of recipients, while alt denotes the proportion of non-directed donors, and instance is simply a marker used to differentiate between the 30 different random instances created for each selection of parameters. Our example file above then has 50 recipients, approximately 5 non-directed donors, and is the 5th instance with these properties generated. The JSON file contains an associative array called "data", which contains one entry for each donor. The key for each entry is the ID of the recipient, and the contents is an associative array that represents the particulars of the recipient. These details include * "sources" : an array of the IDs of the recipients associated with this donor (note that for these files, this array might not exist (if the donor is a non-directed donor), but otherwise will always have length 1), * "altruistic" : a boolean that is true if this donor is a non-directed donor, * "dage" : the age of the donor, * "bloodgroup" : an array mapping "type" to the blood group of the donor, and * "matches" : an array of associative arrays, each of which contains a "recipient" ID and a score assocated with the current donor donating to this recipient. The JSON file also contains a "patients" associative array, describing the recipients. Again, each key is the ID of the recipient, and the value is another associative array containing * "bloodgroup": the blood group of the recipient, * "waiting_time": how long the given recipient has been in the system, in days, * "pra": the PRA level of the recipient, and * "highly_sensitiesd": a boolean denoting whether this recipient is highly sensitised.