Have you considered contacting your data source regarding this? XML parsers are built to be extremely strict and anal about their input to avoid turning XML into what HTML is today: ubiquitous malformed documents littering the web with user agents all interpreting bad mark-up in their own ways. If an XML parser is telling you the data is invalid/malformed, perhaps try it in a few parsers. If you're consistently getting errors, I would try to get the authors of the content to fix the problem.