The Biochemical Journal has started a semantic PDF reader project, which actually has some rather interesting features for scientists, for example converting tabular data to graphs, integrated pubmed search etc.

http://getutopia.com/documents/

The program exists for Mac, Windows and Linux:
There are ubuntu-type installers available as beta and I will attempt making AURs of those.

It would definitely be cool if some of those things could get into Okular or at least that some of the ideas might be an inspiration for further development.