Hi
in the ArTe (www.artentnu.com) project we have been using scratch for dissemination and research goals and we love it.
We have now started to look at scratch programs from a software engineering point of view trying to see if it is possible to define quality metrics for scratch programs.
See http://artentnu.wordpress.com/2010/12/17/how-can-we-assess-the-quality-of-scratch-projects/
At this stage I wonder if there exist a possibility, given a scratch program, to generate statistics about its code (like number of control statements, number of variables, number of sound pieces, etc.).
I also wonder if there are related attempts to define the quality of scratch programs. I am aware that quality is a relative attribute that must be defined according to a goal, but given 1.5 M scratch programs available it would be great if we had a way to select the best ones for teaching purposes, the best for visual aestetics, etc.
Regards,
Ivete
Hi Ivete,
Sorry for not getting to this sooner!
There's a group at The College of New Jersey that you might be interested in. They have a project called "Scrape", which is one way of responding to your question about analyzing the content of Scratch projects.
I think that the project, which is in private beta at the moment, will be presented at SIGCSE next week.
But they have a video on YouTube that explains the thinking behind Scrape.
Thanks,
K