Issue 570: FOL statements in prose -appropriate section of class/property definitions

ID: 
570
Starting Date: 
2021-11-12
Working Group: 
3
Status: 
Open
Background: 

In the 51st CIDOC CRM & 44th FRBRoo SIG meeting, the SIG resolved to work out a well-formulated proposal for where the text rendering FOL statements should be put in the definitions of classes and properties. Probably at the end of the text, right before the examples. 

HW: MD to provide a reading guide for the FOL and where it goes on scope notes. GH, TV, CEO to proofread.

Comments on how to read the FOL statements: 
(i) maybe the labels of classes/properties should appear as well 
(ii) maybe we could draft graphs for the FOL statements.

Current Proposal: 

Post by Martin Doerr (21 January 2022)

Dear All,

Here some attempts to render FOL into language...

I think colors are useful to show which parts render syntax and which variable elements.

 I try to show the transition from literally reading to use of linguistic relative phrases, "which", "this", etc. by offering multiple ways of reading.

Graphical rendering would also be good for the more complex inferences.

Please comment.

Cheers,

Martin

Post by Martin Doerr (23 January 2022) 

Attached a more verbose introduction.

Post by Thanasis Velios (23 January 2022) 

 

Dear Martin,

Please find attached some comments / changes for discussion. I was imagining this as an automated process based on a translation table like e.g.:

∃ "there is"
⋀ "and"
⋁ "or"

etc.

to allow the FOL statement to be translated automatically into English. But maybe this is not possible if we need even marginal elegance in the resulting text.

All the best,

Thanasis

Post by Christian-Emil Ore (23 January 2022)

Hmm, 

The first two FOL clauses for properties are always the domain and co-domain (range). 

It may be considered overkill to repeat the domain-range in plain words. If we should do that, then we could write one sentence:
If x P?? y then x is an instance of E?? and y is an instance of E??'
 

CHr-Emil

In the 52nd CIDOC CRM & 45th FRBRoo SIG meeting, the SIG reviewed the reading guide to FOL expressions (HW by MD). The comments TV had made on the document have not been discussed.

Discussion points: 

 

  • The verb “must” found in the statements is evocative of the necessity operator. But in most cases the relation expressed is that of subset, thus rendering the “must” superfluous (if not ambiguous btw a deontic and a possible interpretation)
  • If “must” is to appear in the textual descriptions of the FOL statements, it should be stated in an introductory text that “must” is to be read as “it is necessarily the case that p” (for whatever lies in its scope). 
  • The material implications apply to every instance of a class or property, not some. This is lost by not introducing the universal quantifier to its reading.
  • However the "for all" or "each" expressions in prose make the statement harder to read for those with no FOL training. 
  • "Must" has been used as a point of emphasis, not to introduce possible world semantics. 

Decision: 
Formulate a proposal re. 
(i) the form of the textual renditions of the FOL expressions
(ii) the position of the rendition  in prose of the FOL in the main body of the text vs. another document/appendix/reading guideline in the intro.

HW: GB, TV, MD and CM to help re-express the FOL statements in English without introducing any sort of ambiguity.

February, 2022

Post by Martin Doerr (7 July 2022)

Dear All,

Attached my proposal to expand FOL statements into a more non-expert-readable form.

I provide more verbous, precise version and a shorter version that reads more fluently. I favour the shorter version. Please suggest better variants along these lines, if you can think of ones. Please comment on possible highlighting with colours, as I have done, separately from comments on the stereotype wording and syntax.

Best,

Martin
    

In the 57th CIDOC CRM & 50th FRBR/LRMoo SIG Meeting, the SIG reviewed the HW that MD had prepared: a guide on

  • the color-coding & fonts used to represent logical constants, variables &quantifiers mentioned throughout the FOL expressions,
  • the stereotypical expressions by which said FOL axioms are rendered in prose. 

For the details of the proposal and the points that came up for discussion see here. 

Summary of decisions: 

  • The shorter versions are preferred over the “standard, complete” versions
  • The domain-range statements are excluded from the textual renditions of property FOLs
  • The textual renditions of FOL statements are to be put right below the FOL expressions in the definitions. 

 

HW assignments:

  • MD to finish drafting the textual renditions to appear in the document and share them with the SIG to be reviewed (and voted upon). Any suggestions for improvements should be given then. 
  • SdS to update the text in the introduction with a “how to read the FOL renditions in prose” section, stemming from MDs guideline. 
  • TV to update the template (specify fonts, indentation, etc.) for textual renditions of FOL expressions. The decision to inform Issue 644. 

 

Marseille, October 2023
 

Reference to Issues:

644
Update CRM templates -shortcuts