TEI/ISO Joint Activity on Feature Structures: Meeting held 24-26 Aug 2004 at AFNOR, Paris.

Report of the project leader

TEI-ISO joint activity

This is a joint activity group formed 2 years ago since the TEI effort in P5 is similar to he ISO one. The project is basically to revise P4.

Plan is to submit of the final document to ISO CS for the approval of the document as a DIS.

October: initiation of the vote for the approval of the doc as DIS

November, submission of the ISO NWIP 24610 FDIS for approval

Joint editorial committee: Kiyong Lee and Lou Burnard: This did not take place.

Side discussion initiated by PD asking about clarifications about the status of the document.

Chairman explains: March 2003 was NP vote. FSR was approved as CD

In November 2003 FSR was approved as DIS from the voting point of view but not approved by ISO CS because the group has not produced the document in the ISO format. This will be done in the near future.

Discussion on comments on FSR

Comments being distributed today are editorial comments on a draft that was not sent to the national bodies... The document distributed is also very recent and is dated of the day of the meeting.

Introduction

Reference to TEI's P5 was changed to P4.

Normative Reference

Many document were removed from the list because they are not used in the body of the document.

Reference to the TEI guidelines should be added.

Terms and definition

LG: Some definitions are not ISO conformant. There is a well-defined structure for definition in ISO documents. Most of the definitions are not formed as definition but explanations.

The standard usually includes the list of terms that need to be understood in order to be able to read document.

Question is can we remove this section and explain concepts differently?

Agreement is that definitions should either be discarded or reworded according to the terminology principles of ISO. The basic scope of the section is fine.

LB will be revising this section.

An example is given that includes a section which includes fundamental notions. Such an approach could be adopted.

Section 4.1 (overview)

This section should be reviewed. What needs to be done is to make clear the relationship between FSR and FSD. This information could be moved to the scope or the introduction. The description which already is in the Introduction has been reworded. Section 4.1 is also rephrased.

Footnote is removed.

Section 4.3

Definition of feature structure is modified to include a mention of collections.

Section 4.4.3

Main comment is that the section does not state clearly that XML is the notation prescribed by the norm. The alternative notation could somehow be confusing. Wording could be made more explicit to show the direct link between graph-AVM-XML. Section on notation is modified to reflect this fact. It should be made clear that example 6, 8 and 15 are equivalent.

Decision is that some XML examples should be kept and directly linked to the examples in this section. Consensus seems to see chapter 4 as an introduction/tutorial, introduction to the chapter should be made more precise in order to reflect that.

EC will go over parts of XML (4.4.3) and modify them. EC suggested removing some examples from the structure sharing section (4.5.1).

Section 4.5

In 4.5, the structure that will be shared should be stated and the illustrated using the various notation. This section has been fixed.

Underspecifiedvalues seem to be causing problems... Section is reworded to make reading easier and 4.5.1 and 4.5.2 are removed.

Section 4.6

Footnote can be removed. Terminology section already mentions that multiset and bag are synonyms. Second sentence is rewritten. Definitions (bag, set, multiset, etc.) found in 5.7 are moved to section 4.6.

A paragraph is added in order to illustrated the fact that collections are harder to illustrate using graph notation so examples are given in AVM and XML.

EC is going to add XML examples for AVMs.

Section 4.6.1

Example (28) in 4.6.1 is modified slightly to make sure the text and the figure correspond. An XML example will be added. Example (29) is complex and does not add much to the document so it is removed. Footnote refering to Shieber is removed.

Section 4.6.2

Example 32 is replaced with a set version of 28. Last part of 4.6.2 was edited/removed.

A note is inserted to underline that, unlike lists, multisets cannot be defined recursively.

Section 4.6.3

Example 34 will be removed since (35) is both simpler and more convincing. The introduction of the section will include a mention of the fact that multisets have not been implemented yet and are included for the sake of completeness.

Section 4.7.1

Section is expanded to include description of how features are selected or put together to create a type.

Formal definition (4.7.2)

Based on TT's comment, the whole section is removed from document

Section 4.7.3

Several minor changes are made to the paragraph.

Section 4.8

Suggestion is to move 4.8 into an annex (EC); this section might be removed in the future.

Section 4.9.1

Delete mention of type from 4.9.1

Remove condition C and insert note mentions that we need to add condition [C] for typed structures.

Section 4.9.3

Example (55) now matches the description included in the document.

Section 4.10.1

Diagram uses a different notation and there is no text to explain the diagram. Decision is to remove this diagram and reference to in the document [KL].

Section 4.10.2

Footnote 21 might need to be removed. Any mention of type hierarchies / constraints will need to be reviewed / removed.

Section 4.10.4

Section was added based on previous comments by participants question its appropriateness. Since this concept is not mentioned anywhere else in the document

Section 4.11.1

Suggestion is to change the title to “Concatenation and Union”. Text is adjusted so that a reference is included to the union operation on "sets” and “bags”. A mention of the fact that the XML notation uses the same representation for both concatenation and union. An example of union will also be included.

Sections 4.11.2

This section on alternation has been redrafted. The last sentence is rephrased to indicate that the use is curly brackets is better avoided. It is suggested to add an XML notation for example (67).

Footnote on page 48 does not use the right char. This will be adopted to use the \vee symbol. Since the note causes confusion, the whole footnote will be rewritten [KL],

Section 4.11.3

This section on negation has been redrafted

Section 4.12

This section on “Informal Semantics” of feature structures is a new draft by HB. Section will be reworked with the help of EC based on a document he wrote. This document will be included in an annex.

Example 70 will be reworked so as to include recursion.

Metamodel (diagram 2) might be removed since HB is not sure of it appropriateness.

References will be taken out.

Section 5.4

Type will be replaced by “label” or something else in the section so as to avoid confusion [LB].

The last paragraph needs to be fixed. <fs> should read <f>. The next example should be adapted so that “fval” is taken out and replaced by “name”. fval's semantic is limited to features, an new attribute should be added to handle feature structures [LB+EC].

Section 5.5

Consistency of the terms will be addressed throughout the section. Last comment should read “... is not recommended for simple atomic values”...

Section 5.6

Example at the end should be modified based in discussion on fval and f.

Section 5.7

Definition of collection should be modified to include fs or atomic values. Collection's definition should also include collections.

The fVal attribute for collection should be removed.

The use of feats on page 65 is wrong since it is supposed to hold pairs (features) and not structures... In this example they refer to feature structures and not features. Basically, feats cannot be used inside coll. Example is also wrong since it contains <f> and not <fs>. Example will be fixed.

Since we might want to use fs contained in a library, we need to add an attribute similar to copyOf in TEI (the fval attribute is kept for f). This needs to be introduced in the document.

Section 5.8.2

EC's comment will be included.

Comment from Takashi Ninomiya will be taken into account.

Section 5.8.3

The second sentence of the section will be modified to put some emphasis on the fact the resulting order can / will be different. [LB]

An example of vcoll for merging or concatenation will be included in chapter 4.

Example at the end of the section is wrong or the text is wrong; this will be fixed to make sure they match [LB].

Section 5.9

Word uncertainty will be removed from the title of the section

Annex A

LB will review / rework it.

KL asks for clarifications about rng and some of the tags used in this annex.

Last page will be fixed so as not to refer to TEI P5 since the document does not exist yet.

The XML example will be removed and replaced with compact notation.

Annex B

TE, EC and KL will review / rework it.

Examples of this annex will be reworked and this section will be merged with annex C.

Annex C

See Annex B.

Annex D

Model theoretical Semantics of FS (EC)

Annex E

Title “Some implementations and applications of FS” (KH), this is current annex C.

Bibliography

Major spring cleaning is done. Only relevant references are kept.

Resolutions

Side discussion initiated by PD

There could be a problem with a 1-month cycle for the approval of the document. KL will check with ISO's officials to make sure this is both possible and accessible.