The machine does the EAV thing. Here is, what I think, is the ultimate strategy. Remember the problem, an application wants from 38 filled with matching members from the entity Joe, from his name to his car type. My graph submits both the form ID and the key words to match and collect.
Now, this pattern is known to the applications, they get it, submit the long form for parts inventory, for example, and just ignore the null value fields. The machine has the Dublin core embedded, it can transform any of the Dublins from one form to another, within their set. The machine gets it, fill in the form, mostly by matching consecutive attribute types in each graph, data and schema. Really just a rewrite of triple key values, maybe allowing some variable operations on them. The result is a set f triples, waddya know!
Where do we get these forms? Look at the web, page layouts, key word sequences in science abstracts, anything that looks like a human nest. This has been done right? Weren't and aren't schema fill in part of c++? Templates anyone?
No comments:
Post a Comment