In the chapter 5 section by Goodchild, he goes over how a geolibrary would work conceptually. However, I think there are some problems already brought up in his description. I don’t think that having server catalogues where a server is in the possession of one (or multiple) specialised collection is necessarily a good idea. It doesn’t sound very efficient to me. The thing about data is that there isn’t equal demand for it everywhere. Some data is demanded more than others (more people search for the weather at a given location than say, the demographic composition of it). Therefore, if you restrict servers to having only a certain kind of specialised content, you would not be optimising your server loads. Instead, you’d get a bunch of servers that have very low traffic that isn’t cost effective, and maybe more servers for which traffic is so high that you need to expand them. I just got the feeling that the suggestion was to model the hardware layout after the layout of the data, but this is just going to be inefficient. It means more connections have to be made, maybe more servers (and the extra costs associated with that), and probably slow performance.
The kind of system being described doesn’t sound very future-proofed. It’s like our ontology discussion. What happens when a new category is created? What happens when a sub-category becomes more important or separated from it’s original category?
In the article on fuzzy spatial queries, I think it is easy to say that we need to be able to form a method of querying that can incorporate both defined an badly defined regions, but how could this ever be compatible with the chapter 5 description of how the library would actually work? I think that ill-defined regions are just that, and I think the best we can do is just use a search engine. If we try to structure it (as we must do if we want specialised servers), then we run into all sorts of problems I think you all have an idea about.
Finally, I’m getting the feeling that we are meant to start our queries in a geolibrary with a location (well defined or not). It seems that geolibraries would be tailored for a certain flow of querying (location > topic > sub-topic > person etc.). What if we don’t want to start our query with a location?
He’s talking about AltaVista….AltaVista people. Remember them? No? That’s because they’re dead.
-Peck
So I take it you don’t have AltaVista set as your default search engine? I just checked to see if it actually still exists – it does (as part of Yahoo). I tried using it to search for McGill University, and the first result was Carleton.ca…whoops!
Good points about the flow of queries – very true that location for many would not be query #1.