On Tuesday, October 2, 2012 10:17:03 AM UTC-4, singhals wrote: > The obvious problem with creating hierarchies is -- who gets > the final say on what goes in? Experts and historians do this work. Whoever provides the data. If it's full of errors we turn somewhere else. We depend upon the knowledge of experts for every aspect of our lives. > Then since clearly hierarchy means something different to > each of us, does your hierarchy begin in the Milky Way > Galaxy and work its fumbling way down to 123 West Main > Street or does it go the other direction? In either event, > does a church parish out-rank a civil parish (where both > exist -- or, I guess, at all). If you'd read what I wrote and understood it, you would realize we are talking about a forest of hierarchies, a directed graph of place names that can be used to handle multi-dimensions of time periods, languages, ecclesiastical dimensions, political dimensions and so on. Do you know what a graph is? This the basic mathematical underpinnings needed to provide the modeling for these services, and it is among the simplest of mathematical concepts to implement as a data structure and write algorithms for. It is the graph structures that allow multiple pure containment hierarchies to coexist. The parallel hierarchies have no "ranking" with respect to each other. They are different hierarchies. This is a subtlety you need to grasp before you can comment meaningfully. > And, in either event of either event -- how will the > hierarchy indicate to the reader that "Glaxon Parish" as a > place-name is incomplete, or whether said parish is civil or > church? Uh, by indicating it. I covered this in what I wrote before. You enter a partial place name. You do or do not try to limit that partial place name to a time, language, region, religion, etc. You push a button. Out comes all the fully qualified names from all the hierarchies in all the dimensions you are interested in. > And yeah, we're in the wet weeds here, but these points have > to be considered. IME, the best time to find out about > unsolvable problems is before you commit significant time, > effort, and/or money. > Cheryl It's not an unsolvable problem. The algorithms are simple. The issue is the creation of the multi- dimensional gazetteer, which is a massive data gathering task. I admit that that task, to be done completely, is a big job. But it can be addressed piecemeal by experts in different areas. The gazetteer can grow and expand over the time. All the "place authorities" that exist today can be thought of as more-or-less one-dimensional gazetteers. In my previous post I indicated that I had written software for the key "graph-node" data structure for use in a truly multi-dimensional place authority. This is not rocket science. There are plenty of examples of how these gazetteers can be formatted. In a future I may try to provide a few examples. Tom Wetmore