You are right about papers. But someone has posted a paper from Stanford about Automatic Generation of High-Coverage Tests and I was so glad to see that.