On 11/10/14 10:52 AM, Simeon Warner wrote:
> On 11/10/14, 1:13 PM, Karen Coyle wrote:
>> On 11/10/14 7:21 AM, Robert Sanderson wrote:
>>> Yes, precisely. The constraints are a distraction that came in from
>>> the discussion over whether it was a good idea to *require*
>>> inferencing or not.
>> But you must address the converse: whether you can forbid or ignore the
>> inferencing inherent in types. In other words, if you aren't supporting
>> inferencing, why are you using types?
> I disagree in the sense that we shouldn't (can't?) try to restrict the
> various ways people might want to interpret the data. We should talk
> about what we are trying to enable instead and be pleased of other
> re-uses that we haven't imagined (key benefits of open data and open
Exactly. As we define types, we need to know that they will be suitable
for the open world and what they enable there. (I recall some
suggestions in this discussion that would have created some very awkward
open-world semantics. We shouldn't do that.) My point is that we can't
just think of the benefit to the folks creating BIBFRAME data, but also
of those who may consuming it, in part or in whole, in the wild. The BF
ontology cannot be limited to one or the other but must serve both. It
is this latter discussion that I have not yet seen happen.
> Perhaps *require* is a confusing word here? It seems important to
> understand the infrastructure necessary in order to effectively use bf
> data. If inference is necessary (avoiding the *r* word here) for
> effective use, then that raises the bar, that has a cost.
[log in to unmask] http://kcoyle.net