Peering into Journal data

Our pilot sites in Jamaica run on the OLPC XS School Server. The XS provides backup services for all the different XO laptops registered to itself. The backup stores the Journal of each laptop. For those who are unfamiliar with the Journal, it is an application that acts as a filing system, except that it does not use the metaphor of files and folders. The Journal stores the child's work and makes it accessible via an interface sortable by time and type of activity.

Sugar Journal

Screenshot: Sugar Journal

Each Journal is backed up to the XS using a random, staggered rsync session. That way, the XS does not get bogged down by a whole bunch of XOs pushing their Journals to it. Once on the XS, the Journal backups can be restored to an XO if it was missing its data or was reflashed recently. The Journal backups are also a repository of the work done by the children. We are able to peer into this repository and extract metadata, so that we can look at aggregate behavior across a classroom or a school. To this end, Leotis Buchanan has written a script that traverses the backup tree and pulls out the Sugar-recorded metadata and dumps it all to a delimited text file.

Screenshot: Listing of the metadata folder

Import this into your favorite spreadsheet software package (such as LibreOffice) or statistical analysis software package (such as R or SOFA) and see what the data tells you. Keep in mind that the data is only as truthful as the applications that create it. Not all Sugar activities report data faithfully, and there are a fair number of missing data holes. These have to be treated in accordance with your analysis methodology. Missing data does not mean it didn't happen. It simply means it wasn't reported.

Screenshot: Data fields


Leotis explains the OLPC XO laptop

Photo: Leotis explains the OLPC laptops and the Sugar learning platrform at the ICT4D Jamaica event.


Many thanks to Leotis for releasing his work under the GNU General Public License. Clone it on GitHub, fork it , improve it and most importantly, reshare it! Leotis has been a part of our community since the very first meeting in September 2008. His contributions continue and are much appreciated. The github repo has the Python script and some sample data from three XOs (XO-1, XO-1.5 and XO-4, approx. 6MB) for you to try out and get started. There is also a CSV file and a corresponding OpenDocument spreadsheet (ODS) file. Upon further consideration, we decided to throw in a XLS file as well for you lesser mortals :-)

Get to GitHub and get going!