Add NBER patent data

[jlcx]

http://www.nber.org/patents/

[jlcx]

The large number of nodes and edges might make this a bit much; any near-term solution here will likely involve some subset of the data. Ideas for interesting subsets:

• all of the predecessors of a patent
• all of the successors of a patent
• all predecessors and successors of a patent (how many nodes could this be?)
• all patents of a category/subcategory/class selected by the user

This raises another question: what tools can be created or utilized to quickly provide pre-calculated layouts of subsets of very large graphs? Or should I say subsets of pre-calculated layouts?