Are you expecting to generate a POJO class without a DTD or XSD? consider the following series of sample "documents".
Based on the first one, you might want getAge to have some numeric return type, but later examples confound that. The first example doesn't demonstrate the non-required sex attribute, and what about allegies? It's only when you get to the third example that you see Person has a list of allegies. This makes me thing that if you wanted to generate a useful POJO from XML documents you would need to supply several example and then use some sort of fancy neural net

algorithms to discover the underlying POJO type. Or bite the bullet and work from an XML Schema
