Identification in reference data libraries

1
Identification in reference data libraries

• Some issues raised by ISO 15926-4

2
What is identification of objects for

• Reference in text
• sentences intended to be read by people that
  refer to objects
• Computer interpretable reference over the Web
• XML name
• Data base ID
• non-readable text string (perhaps just a number)

3
Reference in text
A finned tube air preheater is an air preheater
that contains finned tubes.

• Objects that are defined elsewhere in ISO
  15926-4
• air preheater
• finned tube
• Requirements
• An identifier must be unique within a specified
  namespace.
• That a sequence of characters is an identifier
  must be indicated.
• Where an identifier is an English language
  phrase, modification of an identifier for plurals
  must be allowed in text.
• Where an identifier is an English language
  phrase, modification of an identifier by
  capitalisation of the first letter in a sentence
  must be allowed in text.

4
Reference in text
A PO1395485 is a PO5345 that contains instances
of PO9284.

• This is less useful, so
• we must use words wherever possible
• words must be ambiguous
• The words used must be a unique identifier for an
  object within a namespace

5
Reference in text doing it in plain text
A finned tube air preheater is an air preheater
that has finned tubes.

• One option is
• bold font for identifiers, as in ISO 10303
  standards for entity names
• text at the beginning of the document stating the
  namespace for all words in bold font

6
Reference in text doing it in HTML
A finned tube air preheater is an air preheater
that has finned tubes.

• This is obvious
• use hypertext links to URLs that assign the
  identifiers to defined objects.
• Good to have more semantics than HTML
• XML tagged definition from which HTML can be
  derived
• This is the way the module library works

7
Reference in XML
ltClass IDFinned_tube_air_preheatergt
lttext_idgtfinned tube air preheaterlt/text_idgt
ltsubclassOf resourceAir_preheater/gt
ltdefinitiongtan air preheater that has finned
tubeslt/definitiongt lt/Classgt

• Only one absolute requirement
• an identifier complies with the requirements for
  an XML name
• e.g. no spaces, no silly characters, first
  character not a number

8
Reference in XML
ltClass IDPO1395485gt lttext_idgtfinned tube
air preheaterlt/text_idgt ltsubclassOf
resourcePO5345/gt ltdefinitiongtan air
preheater that has finned tubeslt/definitiongt lt/Cla
ssgt

• Question
• Do we want memorable IDs in the XML?
• It the ability of a person to read the XML a
  benefit?

9
Multiple identification of an object

• We will have multiple identifiers
• A text identifier is not necessarily an XML name.
• We can generate unique memorable XML names
  automatically from the unique text identifiers by
  algorithm.
• A data base identifier is required.
• Equivalent identification
• Each of the identifiers is unique within a
  namespace
• The namespace is usually the standard, e.g. ISO
  15926-4
• Each of the identifiers is normative
• Web identifier of the standard an XML name
  gives globally unique ID
• Identification process
• First unique text identifier
• Generate XML name by algorithm
• Generate data base identifier arbitrarily using
  RDL maintenance system

10
Plural names

• Text identifiers for classes
• will be singular
• will be appropriate to a member of the class
• e.g. finned tube is the identifier of the class
  that contains all finned tubes
• The plurals of text identifiers will be allowed
  in text.

11
Case sensitive names

• XML names are case sensitive
• Will choose to make text names unique
  irrespective of case.
• This will allow capitalisation as required by
  English grammar.
• Can choose a particular uniform case for the XML
  name
• All upper case or all lower case.

12
URNs

• URN consists of
• A prefix identifying the standard (as SC4/N1916)
• The XML name as a suffix

13
Conclusions

• Objects have text identifiers
• Precision over uniqueness and namespace is
  required for a text identifier, as for any other
  identifier
• Objects also have XML names
• Are these derived from the text identifiers?
• Are there deliberately unmemorable XML names as
  well?