Open skybristol opened 4 years ago
Thanks Sky. Adding Dave @dblodgett-usgs to get this thoughts.
I agree @skybristol -- chickens and eggs are everywhere though.
We are ready to start minting IDs at geoconnex.us if we have landing pages to point them at. GNIS, NHD permanent_identifier, NID, HUC codes, etc. etc. are ripe for addition.
I think I've mentioned this before, but maybe not in this context. It's a really good practice to include some type of resolver path mechanism for identifiers to linked information. In the data produced here, you've got identifiers to NHD ("permanent_identifier") and should really have identifiers to GNIS in addition to names. Using a linked open data approach, you would incorporate namespaces into your data model and then reference a base URL for resolving identifiers within those namespaces. Or you could take the expedient of using or including an HTTP-resolvable identifier in your data. This is part of what I think Blodgett and those guys are working to help promote with the geoconnex.us work, standing that up as a registry/resolver/PID infrastructure where needed when a particular system needs a broker or added persistence layer abstracted from the source.
It basically comes down to needing some additional a priori knowledge about how to use the NHD identifiers and GNIS names that your HydroLink system is returning. Those are really meaningful identifiers with a ton of stuff behind them; don't make me have to figure out how to tap into that wealth. If I run this and get the data back, I then need to build in something else outside of that data in order to really use those connections you've helped me create. I'd encourage you to think about the range of next downstream (pun intended) uses that people are going to put this to and try to build in sufficient information that they know exactly what to do next. If you think that additional detail just clutters the data model and it's not useful in your context, then put in a parameter that leaves it out as an option.