- From: Dave Reynolds <dave.e.reynolds@gmail.com>
- Date: Thu, 08 Nov 2012 21:47:28 +0000
- To: public-gld-wg@w3.org
Hi Joćo Paulo, > Will you please point us to examples of usage of the org ontology that we > can use to justify arguments of backward compatibility? It has been in use for UK organogram publication for at least 18 months: http://data.gov.uk/organogram/cabinet-office I believe it has been used by groups in France (who provided the French translation), Spain, and the US (including Health Department information that George mentioned). I don't have public links to any of those examples but Phil was asking for them a few weeks back so may have more. The org:Post extension, while based on the UK organogram extensions [1], was also motivated by requirements identified by Bart van Leeuwen (as part of some EU projects). Bart reported that the extension met those needs. > I am still puzzled with "reportsTo", I've had the chance to look into > this in more detail: > > org:Post is a subclass of org:Organization > > org:Organization is a subclass of foaf:Person > > Thus, org:Post is a subclass of foaf:Agent. Correct. > This means that the union in the domain and range would be unnecessary (as > foaf:Agent would cover all the possibilities). Yes but foaf:Agent is a more general statement than the current union. That generalization wouldn't do any harm but the current approach makes the fact that an organization can hold a post more explicit, which is desirable. > Further, it would mean that org:Organization is within also the domain and > range of reportsTo. In fact, "reportsTo" can be applied to any combination > of foaf:Agent, org:Organization, org:Post, foaf:Group all with different > semantics. No, they are not different semantics. Certainly in normal life people are happy saying they report to a specific person, to a committee or to the holder(s) of post (e.g. "I report to the head of R&D"). I don't think those are fundamentally different semantics of reporting. As we have already agreed, it is certainly possible to have much more fine grained and nuanced representation of reporting but that is out of scope for ORG and should be addressed through extensions (ORG profiles). > I think that saying that org:Post is a subclass of org:Organization could > be part of the problem here. I think we've been round this loop before :) It is the case that in the UK government some posts be held by multiple people at the same time. That does seem to make such Posts a kind of organization (an organized group of people cooperating for a common purpose). That was the view taken in the original UK modelling [1] and I was keen to make the org:Post extension compatible with that since that usage has been reasonably successful. > It leads to other questions such as: what is > the semantics of "subOrganizationOf" when used between posts? Exactly what it would normally mean. If PostA is a subOrganization of PostB then that would imply all the holders of PostA necessarily hold PostB. I wouldn't expect anyone to use such constructs but it does no harm that they are possible. > And > what about foaf:birthday (a property of foaf:Agent) This would also be an > unusual > property of an org:Post. True but just because a property exists doesn't mean it can be applied blindly to any particular instance. There is no cardinality constraint that makes foaf:birthday a required property. Equally I wouldn't use foaf:birthday on an org:Organization even though that is not explicitly banned. > By the way, there are several problems with the non-normative picture in > the working draft. Please be more explicit. If there are specific problems then describe them precisely so they can be fixed. Especially at this very late stage in the process such general statements are not helpful. > Including the omission of "specialization" between > foaf:Agent, org:Organization and org:Post. The diagram is not complete, and not intended to be, some links have been omitted for clarity. That's one of the reasons the diagram is not normative. >>> The semantics of "reportTo" between foaf:Agents >>> is that the agents have some kind of reporting relationship in the >>> scope of some (undetermined) organization. In this case, it would be >>> unreasonable to expect that "reportsTo" is acyclic between >>> foaf:Agents... >>> In contrast, the semantics of "reportsTo" between foaf:Posts is that >>> there exists some kind of reporting relation in the scope of a >>> specific organization. In this case, "reportsTo" may as well be >>> acyclic, as suggested (?) in the current document in a usage note for >>> "headOf". >> >>That's a different issue. It is possible that the comment on acyclic >>relationships is too strong. > > Ok, so how should this be processed into the current draft? (Sorry for the > question, I am new to this!) do we track these issues somehow? If the working group agrees that that is an issue that should be addressed then it gets logged on the issue tracker [2]. That would be a small editorial change and would not affect Last Call so I'd be happy to see that logged as an issue. >>The harm in removing it is lack of backward compatibility and the fact >>that people sometimes just want to say something simple like "fred >>reports to jim". >> >>I've some sympathy with what you are saying here, but am not yet >>convinced the change is justified. Maybe we should see what feedback we >>get from Last Call? >> >>Dave > > I still need to understand the implications of the "Last Call" status... It means the working group thinks the work has been finished and has put it out for final community review. If a problem is identified by that community review then the group has to decide whether it needs to be addressed. If so the spec needs to be revised which may or may not mean a further "Last Call" period. The group can decide that some issues should not be addressed or should be postponed to a future working group and can try to proceed to Candidate Recommendation (CR) phase anyway. The key is that in going to CR the working group has to explain what it has done about all of the issues raised by the Last Call review, one way or another. It is also possible that the work needed to address Last Call comments is more than the time available from the editors or the working group charter in which case the spec is abandoned. Dave [1] http://reference.data.gov.uk/def/central-government [2] http://www.w3.org/2011/gld/track/products/5
Received on Thursday, 8 November 2012 21:47:58 UTC