paper for supersecondary structure prediction

I think this one is pretty similar to our's.

http://contact.ics.uci.edu/betasheet.pdf

I've tried the service BETApro
but the accuracy is about half of it.
so, even combine with our rules, it's hard to get the correct motif.