- From: John Erickson <olyerickson@gmail.com>
- Date: Fri, 9 Sep 2011 14:17:29 -0400
- To: public-gld-wg@w3.org
Thank you very much to those who have responded (thus far...) to my question. A couple notes: Martin asks: > why is marking a datatset as 'official' so important out of your opinion - and/or why for DCAT - as there is for sure information about the publisher & a license given within the metadata - this should be enough basic provenance information (for sure it would be better to have comprehensive provenance info here) to find out / evaluate e.g. the quality of a dataset et al....furthermore by mapping / linking the publisher info with a publisher directory of another source that includes 'official status information' could solve the problem * Having unambiguous URIs for (a) the dct:publisher and (b) dct:license might be a start, *if* we could rely on (in the case of (a)) registries that classified publishers and (b) on licenses expressed as Linked Data, and which actually contained the answer. * A use case might be a harvester that is de-referencing catalog URIs (as if they existed...) and dataset URIs for the purposes of aggregation. * With so many ways to record provenance, and many providers *not* doing so, I'm concerned about relying on inference based on these records to determine the "official" status of datasets * Another way to answer the question, or an addition piece of information, might be to indicate whether the catalog and/or dataset is "Authoritative." <http://data.soton.ac.uk> usefully does this; as I understand it, their working definition is whether the dataset is an ad hoc creation or a by-product of an official system. You can get into nit-picking; for example, catalog and dataset metadata scraped (non-authoritative) from a government site (authoritative, official) because they don't provide the metadata... John David notes: > Since the UN can publish drafts as well as certified datasets, it seems like this requires at least a classification of organizations that publish linked data and a classification of individual datasets, and perhaps a third being the classification of catalogs themselves although not sure how useful that is unless some aggregators are not trusted... * So I think this sort of "certification" of at least publishers/providers would work for particular kinds of certification --- some office of the UN denoting a country's "official" provider. One could even imagine how delegation would work. * David's proposed system of "levels of approval" would work, esp. if the relying service checked where assertions came from * To be consistent with the Open World assumption, there does also need to be a way that status can be expressed in a de-centralized way. It's up to the policies governing the relying system how to use the assertions that it finds... John On Fri, Sep 9, 2011 at 6:33 AM, David Price <dprice@topquadrant.com> wrote: > Some organizations have more authority than others - so UK Home Office might > publish linked data, but so might ISO or the UN or TopQuadrant. Some > datasets have different levels of 'official-dom' - drafts vs. recommended > for certain uses vs. certified as accurate and complete. Since the UN can > publish drafts as well as certified datasets, it seems like this requires at > least a classification of organizations that publish linked data and a > classification of individual datasets, and perhaps a third being the > classification of catalogs themselves although not sure how useful that is > unless some aggregators are not trusted. > > At least one use case I've seen is that in some large organizations, when > starting a new programme they select resources to use based on a preferred > sequence of authorities and levels of approval (i.e. ISO International > Standards, and if not available W3C Recommendations, and if not available > ISO Technical Specifications, and if not available UK government agencies, > and if not available ...). I know this use case is applicable to > organizations as diverse as ISO in deciding normative references when making > standards and in US DOD when approving resources for a new equipment or > research programme. > > Cheers, > David > > On 9/8/2011 10:33 AM, Richard Cyganiak wrote: >> >> On 6 Sep 2011, at 21:27, John Erickson wrote: >>> >>> Questions have arisen as to how to indicate the "official" status of a >>> catalog and/or individual dataset. For example, there are a large >>> number of datasets that are the only source of data for a country but >>> are "Non-government." No properties in DCAT [1] or our own prototype >>> [2] express this adequately. This is important because consumers of >>> catalog metadata must be able to determine whether a source has >>> official status or not... >> >> You use scare quotes around the words “official” and “non-government”. >> >> Can you give a better definition of the distinction you're drawing? >> >> What's the use case for this? >> >> Best, >> Richard >> > > > -- > Managing Director and Consultant > TopQuadrant Limited. Registered in England No. 05614307 > UK +44 7788 561308 > US +1 336-283-0606 > > > > > > -- John S. Erickson, Ph.D. Dir, Web Science Ops, Tetherless World Constellation (RPI) <http://tw.rpi.edu> olyerickson@gmail.com Twitter: @olyerickson Skype: @olyerickson
Received on Friday, 9 September 2011 18:17:57 UTC