I don't see how someone could, in the abstract, determine a function or something that would tell someone the minimal set of variables to put in a DAG.
That is not the problem that this paper tries to solve. The paper assumes you know the graph, and are trying to find a sufficient set of variables to condition on to get d-separation.
To determine the minimal set of variables to include in the graph, you generally need subject matter expertise, ie external causal knowledge. Essentially, you need to be able to claim that there does not exist a variable not in the graph which is a common cause of two variables that are in the graph. (With a faithfulness assumption you may also be able to remove certain variables based on the data)
Over the last year, VincentYu, gwern and others have provided many papers for the LessWrong community (87% success rate in 2012) through previous help desk threads. We originally intended to provide editing, research and general troubleshooting help, but article downloads are by far the most requested service.
If you're doing a LessWrong relevant project we want to help you. If you need help accessing a journal article or academic book chapter, we can get it for you. If you need some research or writing help, we can help there too.
Turnaround times for articles published in the last 20 years or so is usually less than a day. Older articles often take a couple days.
Please make new article requests in the comment section of this thread.
If you would like to help out with finding papers, please monitor this thread for requests. If you want to monitor via RSS like I do, many RSS readers will give you the comment feed if you give it the URL for this thread (or use this link directly).
If you have some special skills you want to volunteer, mention them in the comment section.