This article got published on developerWorks last month:

http://www.ibm.com/developerworks/xml/library/x-DITAdoclet/

This should be of interest to anyone who has generated Javadoc in the past but wished they could get a more semantic, reprocessable format output instead. The doclet produces content using the DITA API specialization, which can be processed with the DITA Open Toolkit. Note that it includes the actual doclet code in the resources section, so you can try this yourself at home.

Comments/feedback welcome, and congrats to Mariana for developing and sharing her process.

--MP