gtm level 1 proposal · 2012-01-30 · gtm level 1 proposal • this is a first gtm level 1...
TRANSCRIPT
![Page 1: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/1.jpg)
http://www.isotopicmaps.org slide 1
GTM Level 1 ProposalJuly 2, 2007
![Page 2: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/2.jpg)
http://www.isotopicmaps.org slide 2
GTM Level 1 Proposal
• This is a first GTM level 1 proposal– intended as a strawman to kick-start discussion– will be properly formalized once feedback indicates that the basic form of the
proposal is accepted by the committee
• Feedback wanted!– is this headed in the right direction?– what is good?– what is bad?– what is missing?– what is too much?– what is not clear?– ...
![Page 3: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/3.jpg)
http://www.isotopicmaps.org slide 3
Mapping to TMCL
• GTM level 1 will have a defined mapping to TMCL– this mapping is not fully defined in this proposal yet
• Note: TMCL schemas are expressed as topic maps
![Page 4: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/4.jpg)
http://www.isotopicmaps.org slide 4
Topic types
• Topic types are always boxes• A QName (or id) giving the
subject (or item) identifier mustbe present
• Prefixes are declared withfloating text in CTM syntax
foo:person
%prefix foo http://psi.example.org/
![Page 5: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/5.jpg)
http://www.isotopicmaps.org slide 5
Properties
• A division for names may ormay not be present
• A division for occurrences mayor may not be present– must always be the second
division– first can be empty
• Cardinality is omissible• Datatypes are omissible• @ oasis:language means
foo:biography can be scopedwith topics of this type
foo:person
tm:name 1..1foo:given-name 1..1foo:family-name 1..1foo:email : string 1..1foo:biography : uri 0..* @ oasis:language
![Page 6: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/6.jpg)
http://www.isotopicmaps.org slide 6
Abstract topic types
• As in UML– write the topic type identifier in
italicsfoo:person
tm:name 1..1foo:given-name 1..1foo:family-name 1..1foo:email : string 1..1foo:biography : uri 0..* @ oasis:biography
![Page 7: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/7.jpg)
http://www.isotopicmaps.org slide 7
Binary associations
• Binary associations are lines– association type given in middle– role types given near player– cardinality from player side given
near player
foo:person
tm:name 1..1foo:given-name 1..1foo:family-name 1..1foo:email : string 1..1foo:biography : uri 0..* @ oasis:biography
foo:organization
tm:name 1..1
foo:homepage : uri 0..1
foo:employed-by
foo:employee
foo:employer
1..1
0..*
![Page 8: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/8.jpg)
http://www.isotopicmaps.org slide 8
N-ary associations
• N-ary associations use a circleto represent the associationtype– behaviour is otherwise as for
binaries
foo:person foo:organization
foo:position
foo:employed-by
foo:employee foo:employer
foo:position
![Page 9: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/9.jpg)
http://www.isotopicmaps.org slide 9
Unary associations
• Unary associations follow thesame pattern
• There must be restrictions onthe possible cardinalities herefoo:person
foo:is-alive
foo:living
![Page 10: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/10.jpg)
http://www.isotopicmaps.org slide 10
Symmetric associations
• Binary associations where thesame role type appears on bothsides
• Issue: cardinality on both sidesmust be consistent
foo:person
foo:friend-of
foo:friend
foo:friend
![Page 11: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/11.jpg)
http://www.isotopicmaps.org slide 11
Repeatable roles
• For n-ary associations thecardinalities of roles in instanceassociations can be given
• For binary associations theyare fixed at 1..1 (except ifrepeated, as in symmetrics)
• Disclaimer: this is not anexample of good modelling
foo:person
foo:soccer-team
foo:coach 0..1
foo:player 0..1
11..*1..1
![Page 12: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/12.jpg)
http://www.isotopicmaps.org slide 12
Roles with many player types
• It’s possible for a role type tobe played by more than onetopic type
• This is represented bybranching the line– works for both n-ary and binary
• Role type given at branchingpoint
• Cardinalities given at player
foo:document
dcc:resource 0..*
foo:person foo:place foo:organization
dcc:value
dc:subject
0..* 0..* 0..*
![Page 13: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/13.jpg)
http://www.isotopicmaps.org slide 13
Subclassing
• Effectively UML notationfoo:person
foo:employee
![Page 14: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/14.jpg)
http://www.isotopicmaps.org slide 14
Text notes
• Text notes documenting thediagrams are allowed
• They are given as simplerectangles containing the textnote
foo:person
foo:employee
We realize that employee is strictly speaking a role type,but...
![Page 15: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/15.jpg)
http://www.isotopicmaps.org slide 15
Identity constraints
• A separate division for these– divisions have a fixed order– names, occurrences, identities
• Predefined names– locator (subject locator)– identifier (subject identifier)– itemid (item identifier)
• Datatypes fixed to “uri”
foo:organization
tm:name 1..1
foo:homepage : uri 0..1
identifier 1..*
![Page 16: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/16.jpg)
http://www.isotopicmaps.org slide 16
Issue: What about names?
• This proposal does not put names for typing topics in the diagram– the rationale is that space savings are crucial for readability in large diagrams
• So where are the names specified?• In this proposal that is considered out of scope
– editing tools can allow the names to be edited manually– and/or they can generate default names from the PSIs– or they can ignore them entirely
![Page 17: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/17.jpg)
http://www.isotopicmaps.org slide 17
Issue: scope support
• The support for scope needs more work• Open questions:
– interaction of cardinality with scope– multiple types of scoping topics– ...
![Page 18: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/18.jpg)
http://www.isotopicmaps.org slide 18
Issue: reification support
• Should there be any?• What should it look like?
![Page 19: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/19.jpg)
http://www.isotopicmaps.org slide 19
Issue: assertion constraints
• Should query constraints be supported?• Should regular expression constraints be supported?
![Page 20: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/20.jpg)
http://www.isotopicmaps.org slide 20
Issue: omitted information
• GTM must indicate which information can be omitted• How should omitted information be interpreted?
– should there be default cardinalities, for example?– is it possible to generate TMCL without specifying these?– is it better to just leave the issue of defaults to tools?
• What about visual shorthands for omitted information?– these would serve as indicators that something is present but not shown– is that useful? is it clutter? is it too much complexity?
![Page 21: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/21.jpg)
http://www.isotopicmaps.org slide 21
Issue: overlapping types
• It is possible for topic types to overlap– for example: in the Italian Opera topic map the librettist and composer types
overlap, in the sense that topics can be instances of both
• In TMCL overlap must be explicitly stated to be allowed• Should GTM support this?
– if so, how?
person
composerlibrettist character
![Page 22: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/22.jpg)
http://www.isotopicmaps.org slide 22
Issue: codependent role player types
• Given an association type contained-in that joins– cities, provinces, and countries, where– cities must be in a province, and provinces must be in a country– it’s not allowed to connect cities directly with countries
• This constraint is expressible in TMCL• Should it be expressible in GTM?• If so, how?
place
city province country
contained-in
containee
container
![Page 23: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/23.jpg)
http://www.isotopicmaps.org slide 23
Issue: support for multiple schemas?
• TMCL allows multiple schemas to be mixed in a single topic map• Should GTM allow diagrams to indicate which schema they “belong
to”?• The current proposal stays well clear of this
![Page 24: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/24.jpg)
http://www.isotopicmaps.org slide 24
Issue: interchange format for graphical info
• We propose that we not support this
![Page 25: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/25.jpg)
http://www.isotopicmaps.org slide 25
Issue: documentation of non-topic types
• Should it be possible to make a GTM diagram that says– dc:description is an occurrence type with datatype string?
• That is, without assigning the occurrence type to any topic type...• This would make it possible to create diagrams for ontology
fragments– on the other hand: is that useful?
• Is the thing on the right the solution?
dc:description : string
![Page 26: GTM Level 1 Proposal · 2012-01-30 · GTM Level 1 Proposal • This is a first GTM level 1 proposal –intended as a strawman to kick-start discussion –will be properly formalized](https://reader033.vdocuments.mx/reader033/viewer/2022050402/5f8070444c91ee2be7523cc8/html5/thumbnails/26.jpg)
http://www.isotopicmaps.org slide 26
Issue: navigable roles
• UML allows navigable roles to be specified– that is, to say that an association is only traversable in one direction
• Should GTM and TMCL support this?– it would not be a hard constraint, but more in the nature of a semantic hint