Root element of the data set.

Information associated with the creation of the dataset in human-readable form. May be augmented or replaced with common TDWG MetaData structure or UBIF container.

Person or Program that generated the data set.

Time at which the data set was created.

Reason why the data set was generated.

Origin of the data for derived data sets (e.g. database connect string).

Mechanism to allow for the extension of the schema for specific applications.

- specific GBIF request

Container for voucher (representation of specimen) records.

Specimen and location [A]

Information about the location of the specimen. The PlaceholderType represents a placeholder for optionally extending the schema with a schema element defined in another standard. [A]

The PlaceholderType represents a placeholder for optionally extending the schema with a schema element defined in another standard. [A]

Physical object the voucher is referring to. The PlaceholderType represents a placeholder for optionally extending the schema with a schema element defined in another standard. [A]

Container for publication records.

Details of the data source where the the concept was originally published. (Might adopt a common TDWG data structure.) 

Full bibliographic reference as a single formatted string.

Reference broken down into individual components. (The current version is based on R.Pyles's Taxonomer data model, which in turn is based on Endnote 7.) [A]

Container for concept records.

Representation of a Taxon Concept. Various types of concept can be represented, including a reference to the GUID of an existing Concept. [A]

A non-unique handle to the concept. [A]

Complete name as found in the data source. May require calculation from an atomized record of a name in the source data.

Atomized data items of the scientific name, if supplied by the source data. This is equivalent to the PlaceholderType mechanism employed for vouchers and publications, but already instatiated with the agreed ABCD standard.

Information about the authorship of this concept which uses the Name in their sense (i.e. secundum, sensu).

Stores explicit, taxonomic and nomenclatural relationships that are part of the original concept definition.

Type of the relationship [A]

Reference to another TaxonConcept. [A]

A set of specimens that are used to define the concept.

Link to record of a specimen. [A]

A set of taxonomic descriptions used to define this concept. May potentially hold descriptions according to the TDWG SDD schema, or any other, format.

Mechanism to allow for the extension of the schema for specific applications.

- specific GBIF request

Container for taxonomic assertion records.

Relationships between two concepts which are not part of the original definition of either of these concepts; possibly by a third party. [A] 

Information about the authorship of the asserted relationship.

Starting point of the directed relationship. A reference to a TaxonConcept.

End point of the directed relationship. A reference to a TaxonConcept.

Contains only the human readable label string (multilingual) and mutliple linking mechanisms (URL, LSID, etc.).

Extensions of 
ProxyBase
specific to 
PublicationProxy

Describes the origin of the concept or assertion.

Unstructured string as used in the data source describing the origin of the concept (e.g. AuthorTeam and year).

Details provided by the data source about the concept or relationship authors.

String representation of  who was responsible for creating a particular concept or relationship, as represented in the source data. May be a subset of the publication authors.

Time of creation of the concept or relationship .

Reference ID or GUID of the original publication in which the concept or relationship was introduced. [A]

Specifies a location (e.g. page or paragraph) within the publication.

A collection (seq) of name strings, used for publication authors or editors and for collectors, i. e. whenever the identity of an Agent is doubtful and can not associated with an Agent without doubt

Authors or Editors expressed only as string, e.g. in publications where the identity of creators can often not be discovered. Optionally, the ref attribute may refer to an agent if the relation between string and Agent can be assessed. (The sequence of elements in instance documents is informative!)
 [ATTR: ref]

The full code-appropriate author team string for this name at this rank. If author is not known, enter the literal "-" (element is not optional!). Examples: 'L.'; '(L.) Smith & Jones ex Brown, Green & Black'. Where atomized citation data exist, this text should be derived from them.

GH: I propose to  require a label containing the canonical name without authors. I wonder whether this should replace the Monomial below, i.e. in the case of Genus and above, Text contains a mononomial and in other cases a canonical concatenation.

SCH (Bowing to GH's request) the family, genus, infrafamilial or suprafamilial name

Flag indicating that this is a named hybrid or a graft chimaera. SCH: should we be nailing this down as either a boolean (Hybrid/ notHybrid) or a controlled string (x, or mult.sign, or empty string (i.e. definitely not a hybrid) or null (i.e. not sure if it's a hybrid or not.) I may have got my nulls and empty strings mixed up here.

GH: I moved this up here, not sure it truly belongs here.
@@Please add annotation

This name is a nothotaxon name

Date separated into attributes so that any part of the date may be missing
[ATTR: year = four digit year;
month = two digit month of year;
day = two digit day of month;
verbatim = unparsed textual date representation;
supplement = text additional or modifying the exact dates, e. g., 'end of summer', 'first half or year', 'first decade of month', '1888-1892';
timezone = expressed as integer according to the xml schema seven parameter model]

The object linking 
mechanisms used 
by the ProxyBase 
type may also be 
used by other 
objects!

LifeScience ID (including the constant prefix 'urn:lsid:', followed by 3 to 4 parts separated by colon). The 1st part is the url of a lsid authority service that provides metadata on how to obtain the object references in part 2 (namespace = data collection), 3 (object ID) and 4 (optional object version). Example: 'urn:lsid:lsid.gbif.org:DataCollectionID:ID/1§31~b+:v2'

Digital object identifier (an ID scheme advanced by the library community).
Example: 10.47198/923347'

A URL directly providing an object representation. In contrast to the URN types LSID or DOI this should resolve directly. The URL may be a query string (with ID embedded), for example: "http://x.y.fr/pub/au=smith?yr=1998".

In the case of URLs multiple definitions may be defined to reduce the likelihood of failure. 
[The element sequence  in instance documents is informative and should be preserved.]

Defines an element with a required 'language' attribute

Gives enough information about a name to unambiguously refer to it

The full name of the linked name. With enough information to be unambiguous. If no LinkId is provided this should/could include author and year. If not, maybe just the name or epithet

Links to the actual record if available

Combines an AlexandriaCore citation with taxonomic authors, location in publication like page, etc.

Taxonomic authors

The problem of ex (I vote to include them where possible) and ":" (sanctioning) require additional structure nested within the Protonym/Combining authors structure.

At the moment I prefer to define lowercase "ex" and ":" as special values for author names, and simply list them in the sequence to reduce the complexity of the model.

Placeholder for other schema standards

Example PublicationType to be used instead of PlaceholderType

SCH: the id attribute contains the locally or globally unique Identifier for this name record. Could be one of three kinds of ID (see attribute "type"): GUID (persistent, machine resolvable such as an LSID), human resolvable Global ID (e.g. 'IPNI citation ID 14098-1') or locally (within this dataset) resolved ID (e.g. '1' or 'K001928712') - probably belongs somewhere else if it looks useful and could end up being widely used 

Reference to an entity defined elsewhere (equivalent to a foreign key)

Type attribute describe the type of relationship

GH: (formerly FullName) A full name, including nomenclatural citation (authors/year in the form mandated by the relevant code). The name should be normalized (removing optional infrageneric or infraspecific hierarchy, author given only for the lowest nomenclatural rank) human readable and unique representation of the nomenclatoral name object. Where available, this should be generated from the canonical name and authorship plus Nameextensions. In combination with the ID this
 constitutes the lowest (level 1) abstraction of LC name. Any other information is optional. 

NomenCode of the concept.

Rank of the concept (potentially enumerated to the UBIF standard). 

SCH: Canonical name enforcing strict inclusion of only nomenclatural information - _not_ taxonomic information with the exception of the necessary placements within Genus or Species

Authorship of the name, optionally atomizated.

@@ GH added for further discussion

Races and possibly other ranks not governed by a code (patho/bio/serovar).

The name of a cultivar group as specified in the ICNCP. Example "Granny Smith Group"

The name of a cultivar as specified in the ICNCP. Examples: Firmgold, Granny Smith. Enter without quotes; single quotes should be added when creating the full name.

Jerry: ###These can occur at any rank and are therefore treated as an extra 'label' attached at any rank. We should develop this more fully. Related issue: Trade Names. These are allocated by a registering authority on a country-basis and thus the same trade name can be applied to different taxa in different countries.

The spelling given to the full taxon name by the author(s) originally creating the name, if not accepted under code. Reasons for later orthography changes include spelling, grammar, and gender corrections, interpretation of undefined ranks like "beta, gamma". Multinomials expressing original infrageneric taxonomic opinion (e. g., 'Saxifraga aizoon var. aizoon subvar. brevifolia f. multicaulis subf. surculosa Engl. & Irmsch.' should also be noted here. AKA OriginalVerbatimOrthographyAsPublished.

The should be identical with AlexandriaCore - currently only temporarily imported.

Human readable representation. This may be the only data item if no machine readable ObjectLink exists. Example for a publication: "Smith 1998. Flora of Erehwon, X Publ." Even if an external ID exist, the Label is required. It preserves the semantics of the proxy object (= keep interpretable by humans) if the machine-readable object links are broken. Label should be updated automatically (without human control) only after a human decided that the semantic management of an external object provider can be fully trusted. In an LSID-based interaction, Label is equivalent to the 'title' metadata.

The provided Abbreviation element is not necessary for all proxies, but especially useful for class names (e. g., for tabular reports) and publication abbreviations (author/year style).

Label text in a specific language.

Restricted to 50 characters maximum length, including blanks (recommended to be shorter!). Label abbreviations are especially important when displaying information in a tabular format.

Defines an ID of an external object or one to several services providing it.

Year as appearing on the publication. Compare TruePublicationDate below.

Effective date of publication; may be different from year stated on/in the publication. Important for taxonomic or other priority.
[ATTR: year, month, day]

Series of books or 
articles (the latter 
may be published 
in edited books. 
journals, or on 
the web).

Series title

Series editors

International Standard Serial Number

Periodical/magazine
/journal information

(should ideally come from BPH, TL2 and other standard dictionaries)

Standardized abbreviated form of title

International Standard Serial Number

Printed book:
monograph, 
thesis, report, 
or edited book 
with articles

Book title (monograph or edited book)

Creators are authors 
if book is a monograph, 
editors if it contains 
articles by other authors

Volume or part number in a series

Total range of pages, including foreword, appendices, index and plates/figures.

International Standard Book Number
(if multiple numbers exist, delimit by ";")

Number of the edition of a book.

Publisher, reprint year, note, etc. for historical books that are reissued.

Publishers of a book, periodical,
or independently published article.

The name of the publisher (publishing company or institution, including universities or scientific societies).

The publication location where the item being referenced was published, such as a city and state.

Articles may, e. g., 
be published in 
periodicals, edited 
books, or on the internet.

Volume of periodical
(empty if article appears in edited book)

Part or issue of a periodical volume
(empty if article appears in edited book)

Pages of article. This may include table, or figure numbers for the reference. Examples:  '23-41', '341 pp.', or '20, 22-24, 32' (for non-consecutive pages).

Optional information about a chapter, 
section, etc. that has the same authors 
as the publication in which it is contained. 
Compare Article for authored chapters 
in edited books.

Number of chapter, section, etc.
as used in the publication.

Pages of current part ('22-34')

String (i. e. xs:string with minimum length=1) extended with *optional* language attribute

String255 (i.e. xs:string with length 1-255), extended with *optional* language attribute

Three attribute provide options to express taxon rank as code (enumerated vocabulary), free-form text (perhaps interpreted), or verbatim (uninterpreted original version). At least one attribute should be present; this can not be validated by the schema.

Text with primary language
plus multiple optional
translations; used, e. g., in
PublicationProxy type.

A string, e. g. the title of a publication, having a single primary language.
[ATTR: language]

Translations from the
primary language

[ATTR: language]

Digital Object Identifier (standalone, not embedded into URI syntax)

Union of xs:language with '-' for language-neutral (e.g. scientific names) and '?' for unknown. Language follows RFC 3066 'Tags for the Identification of Languages': a two-letter code taken from ISO 639 part 1 or a three-letter code taken from ISO 639 part 2, followed optionally by a two-letter country code taken from ISO 3166. (Notes: When a language has both a two-letter and three-letter code, use the two-letter code. RFC 3066 replaces RFC 1766.)

Life Science ID (= string restricted by a regular expression pattern).
Annotation of the pattern: 5 to 6 parts separated by colon
1. The string URN (case-insensitive)
2. The string LSID (case-insensitive)
3. AuthorityID = DNS token with at least 2 parts plus a top-level domain with 2-5 characters (case-insensitive)
(In earlier LSID specs this was assumed to be a DNS name; the final spec. however says: "The authority identification is usually an Internet domain name. In this case it is recommended that it be owned by the organization that assigns an LSID in question. Such organization is responsible for ensuring the uniqueness of the string created from the namespace, object and revision identifications. In the case where the authority identification string is not an Internet domain name, the authority should take care to ensure that it is a unique string and if possible, register that unique string with the organization that is currently the authority for the URN Namespace Identifier (NID) "lsid""
4. Data collection identifier/namespace: non-whitespace characters except colon (case-sensitive)
5. Object ID: non-whitespace characters except colon (case-sensitive)
6. Object version (optional): non-whitespace characters except colon (case-sensitive)

Earlier, more specific specs at http://www.i3c.org/wgr/ta/resources/lsid/docs/LSIDSyntax9-20-02.htm had more restrictions on authority (DNS name!) and fewer
characters beyond US-ASCII and digits. A pattern matching the earlier spec. was (extended with "local"):
pattern value="[Uu][Rr][Nn]:[Ll][Ss][Ii][Dd]:((local)|(([0-9A-Za-z\-]+\.){2,}[A-Za-z]{2,5}))(:[0-9A-Za-z][0-9A-Za-z\(\)\+,\.=;$" _!\*'\-]+){2,3}"

Example: urn:LSID:www.gbif.net:DataCollection.Namespace:ID/$17+731_b:v2.1

Enumeration of the nomenclatural code under which a name is considered valid. (Source: comparison of enumerations in ABCD 1.49 and initial LinneanCore.)

### This needs urgently revision! Enumeration of possible values for nomenclatural status. (Source: initial LinneanCore.)