Im trying my best to catch up with all the mails in this thread lately, but surely missed something. Ive ended up with a rather long post, please excuse, but I think the new dwc draft needs some explanations...
The idea of darwin core is to provide a simple list of terms which are useful in different contexts - just like dublin core does. We have been thinking about giving the terms a "domain", the dublin core terminology for binding a property to a class. Although we did declare a terms domain in the description, we finally decided that this is not more than a grouping of terms. Assigning a domain is a rather controversary task as properties can belong to several classes and the granularity of class definitions is strongly depending on your views. This is likely the mayor obstacle in getting an agreed ontology on the road too I believe.
But in order to express the context of a collection of darwin core terms we also defined class terms representing a taxon/name, an occurrence/specimen, a site, a collecting event and more. Actually there are 2 ways of expressing the type of a thing in the new terms - using a class term (e.g. as the parent xml element instead of just record or inside the rowType attribute of the text archive meta file) or using the basis of record when using the simple/classic darwin core with a meaningfree DarwinRecord parent element.
During last TDWG I got very attracted to the KISS idea that was present all over, but especially in Chucks talk. Why has darwin core been so much more successful than other tdwg formats? Why can't we use the same approach to share taxonomic or nomenclatural data? In the end the core properties for a taxon or name are rather limited and if you take TCS apart there is not much more you can do with it than what the taxonomic dwc terms provide - leaving concept relations aside. We also decided not to separate a name from a taxon. That happens in the context and due to the present of a taxonomic status or taxonomic classification. In order to save you from looking up the latest draft, here is the list of the taxonomic terms with a few quick annoations from myself:
[ taken from http://darwincore.googlecode.com/svn/trunk/terms/ index.htm ]
taxonID # the taxon/name ID, preferrably a GUID but local ids are permitted too. Any ID scientificName # the full name incl authorship higherTaxonID # ID pointing to the next higher taxon higherTaxon # full scientific name of the next higher taxon. Can be used alternatively or in addition to the ID above kingdom # the classic dwc higher ranks to simplify transfer in many cases and useful to disambiguate homonyms phylum class order family genus subgenus specificEpithet # atomised epitheton part as you would guess taxonRank # any rank string, preferrably taken from a controlled list such as the TCS one infraspecificEpithet scientificNameAuthorship # the authorship alone. Similar to the other atomised parts this might not be needed as the complete sciname string is good enough, but it seemed desired by many nomenclaturalCode # the code to disambiguate homonyms across the codes taxonAccordingTo # the taxon concept sec reference namePublishedIn # the nomenclatural reference, original publication of the name taxonomicStatus # a status such as accepted, (heterotypic) synonym, missapplied name. Ideally also a vocabulary such as the ontology one for nomenclatural note types nomenclaturalStatus # similar to above, but only nomenclaturaly relevant statuses such as nom. illeg. acceptedTaxonID # pointer to accepted taxon in case of synonyms or misapplied names acceptedTaxon # explicit string version of above basionymID # pointer to the basionym basionym # explicit string version of above
So frankly I think there is a lot you can do with these terms. Every synonym will be a name on its own and link to the accepted taxon, having its own taxonomic status that can provide you with details about the reason for it being a synonym or misapplied name. If combined with 1 to many extensions, you could even exchange pretty detailed species information in a *very* simple model. Imagine you have datat based on extensions for species such as distribution (multiple distribution status per area per species), descriptions (just a simple title, body, and description type ala SPM classes would be great), multimedia, alternative links/guids, types data. As an exercise I have also translated parts of the GISIN schema into extensions based around species described by darwin core terms and it works fine.
Surely darwin core terms can leave you in doubt about the exact context, just as dublin core does. But it immediately allows you to share data in a simple way that people are comfortable with and it is not tight to a technology per se. I see this as the biggest advantage of all. It plays nice in the world of xml, rdf, plain text, xhtml, microformats - just anything as it is based on just a list of terms with an associated globally unique URI. And by decoupling the recommendations on how to use the dwc terms in the context of different technologies, we can keep the term definitions stable no matter what comes next.
Also I wanted to avoid confusion about IPT and the dwc text archives. The text archive, extensions and vocabularies idea implemented in the IPT are not IPT specific. It's just a very simple exchanging format based on the dwc recommendations for how to use darwin core in the context of text/csv files. For the interested I am writing on client code currently, so there will be a java archive reader in a few days that allows you to easily iterate over the core dwc records in an archive and pulls out the related extension records together with the core record.
I am glad to see this list being so alive again! Markus
On Apr 28, 2009, at 3:22, Blum, Stan wrote:
@font-face { font-family: Tahoma; } @page Section1 {size: 8.5in 11.0in; margin: 1.0in 1.25in 1.0in 1.25in; } P.MsoNormal { FONT- SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } LI.MsoNormal { FONT-SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } DIV.MsoNormal { FONT-SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } A:link { COLOR: blue; TEXT- DECORATION: underline } SPAN.MsoHyperlink { COLOR: blue; TEXT- DECORATION: underline } A:visited { COLOR: blue; TEXT-DECORATION: underline } SPAN.MsoHyperlinkFollowed { COLOR: blue; TEXT- DECORATION: underline } P { FONT-SIZE: 12pt; MARGIN: 0in 0in 0pt; FONT-FAMILY: "Times New Roman" } SPAN.EmailStyle18 { COLOR: navy; FONT-FAMILY: Arial; mso-style-type: personal-reply } DIV.Section1 { page: Section1 } Please note, the most current draft of the DarwinCore is: here http://code.google.com/p/darwincore/ and here http://rs.tdwg.org/dwc/index.htm, not here http://wiki.tdwg.org/twiki/bin/view/DarwinCore/DarwinCoreDraftStandard) ,
My primary concern with the latest draft is the absence of an explicit class identifier (and an implicit class definition) that indicates what kind of data object the sender is transmitting. If an indexer/aggregator is indexing multiple kinds of resources, as GBIF is, and a publisher provides a record with these elements [ScientificName, ScientificNameAuthorship, NamePublishedIn, and Country], how should the indexer interepret this record? Is it an organism occurrence record, an authoritative taxonomic record (the country name indicating the entire known range of the taxon), or part of a taxonomic checklist for that country? The term/element [BasisOfRecord] is the first step in narrowing the possible meanings, but it's the only step and it appears not to be a required step. (I interpret the "Status: recommended" to mean that it's optional.) At a minimum, BasisOfRecord should be required. It would still be possible to publish garbage (at least hard to interpret records) because our tools don't constrain structure, and there isn't (yet) any guidance controlling the structures for different classes of objects.
The new GBIF Internet Publishing Toolkit (IPT) supports one-to-many relationships among a series of flat tables and looks like it's going to make it easier to transmit more complicated data than we were doing with DiGIR and TAPIR. In conjunction with this new expanded bag of elements we could see a lot more complex data get published. If I may use a skiing metaphor, we've been on the bunny slopes (for beginners) up until now. The new DarwinCore looks like a nicely groomed black diamond slope, but we haven't had any lessons or even watched anyone else do what we're going to attempt. I think we're going to end up in a heap at the bottom of the hill, and the aggregators are going to have to sort it out. ( Woohoo! Extreme biodiversity informatics! )
Finally, I did not mean to say that the Ontology should not be ratified, as in not supported; what I meant was that it should not be a standard because we will want to change its contents without versioning the ontology as a whole. (Also, its not an application schema, so it can't be used directly unless we venture into somewhat uncharted territory.) Its role is to help us keep our application schemas coordinated.
The earlier versions of the DarwinCore (or our protocols, or the way we used them together) were too limiting (see Greg's comments); this version allows terms to be combined in nonsensical ways. It could make life very difficult for data integrators.
I think the bottom line is that we URGENTLY need a similar concerted effort to advance the TDWG (biodiversity informatics) ontology, and a companion set of application schemas coming forward from the collections, marine biology, paleo, observation, taxon-name-concept groups as soon as possible.
-Stan
-----Original Message----- From: Chuck Miller [mailto:Chuck.Miller@mobot.org] Sent: Monday, April 27, 2009 7:22 AM To: Kevin Richards; Blum, Stan; Technical Architecture Group mailinglist; exec@tdwg.org Subject: RE: [tdwg-tag] darwin core terms inside tdwg ontology
Kevin,
I agree with you and Stan that the ontology is useful to all schemas. It seems to me that a “TDWG Ontology” is a totally new and different kind of thing than all the data exchange standards of the prior 10 years – DwC, SDD, TCS, etc. But, it is a very useful and important new kind of thing that should be part of the TDWG standards architecture. It challenges prior thinking about the nature of TDWG standards to grasp what standardizing on an ontology means. But, I think it’s what is needed.
If TDWG standardized on one Ontology, then the vocabulary of all data exchange could be standardized on it. Then all TDWG standards could be revised over time to comply to that vocabulary standard, including DwC.
Stan said: “ I'd like to hear the rationale for combining taxonomic name/concept with organism occurrence.” An occurrence record generally has an organism’s name associated with it in the real world. It is necessary and inevitable that vocabulary about organism names will be used in an occurrence data exchange schema like DwC. We have been stymied with this idea for years. A standard Ontology/ vocabulary for the elements of name information needed to be associated with an occurrence, or a description, or a taxon concept would go a long way toward solving this duality. The “standard vocabulary” would not be standardized within DwC but it would be used in DwC.
Of course there is the problem of the hundreds of installations of DiGIR that use DwC “classic” and are no doubt not going to change for a long time. I think they just have to be accepted and worked around going forward. It’s impractical to think of anything else. But, the past should not roadblock the future and we need to get moving toward that future.
Stan thinks that the Ontology is not appropriate for TDWG ratification. Why not? Change has to start somewhere. Yes, other standards would probably be in conflict if the Ontology were ratified, but I think we want to ultimately have consistency across all the standards and that means there has to be change going forward. I think a ratified TDWG Ontology would provide the foundation upon which to start building those changes.
Chuck
tdwg-tag mailing list tdwg-tag@lists.tdwg.org http://lists.tdwg.org/mailman/listinfo/tdwg-tag