This paper argues that it is useful to be able to trade recall for precision in dependency parsing, and suggest some methods of doing so.
I believe the precision biased task is in some ways more interesting than the specific solution proposed in the paper.

The paper was rejected from various conferences, the latest of which was EMNLP 2012.  It was usually the case that one reviewer found the paper to be interesting, while the others just "didn't get it".  

I believe the task and results is intersting, but am bored of revising and resubmiting it.
I decided to publish it on my webpage + arXiv instead (along with the reviews).

If you found it to be useful also, I'd be nice if you drop me a line and/or cite it as a tech-report.