[SCL] Fwd: Common Logic status with respect to standards

[Harry Delugach]

On Friday, October 31, 2003, at 05:11  AM, pat hayes wrote:

>> As many of you know, I've volunteered to serve as editor of the 
>> Common Logic standard that is being proposed to ISO. Having attended 
>> my first ISO meeting, I'd like to get the standards effort organized. 
>> So here are my thoughts. These are purely administrative notes; I 
>> will have a separate technical note in a day or two. Please don't 
>> overlook the last paragraph!
>>
>> The current status of Common Logic is as a Work Item assigned to 
>> ISO/IEC JTC1/SC32 WG2. There is a joint technical committee between 
>> the ISO and the IEC called Joint Technical Committee 1 (there is no 
>> number 2). Under JTC1 there is a subcommittee on information 
>> technology called SC32. Under that subcommittee, there is Working 
>> Group 2. This is the international body responsible for metadata 
>> standards. The steps in the process are briefly as follows:
>>
>> A working Draft (WD) needs to be submitted to WG2 for comments. It 
>> will be called ISO/IEC WD 24707. It must conform to ISO formatting 
>> standards, which basically means using a Word template. In order to 
>> get the process rolling, we will need a working draft for the INCITS 
>> L8 meeting in Washington on 17 Dec 2003.
>
> Suppose we do not make this deadline: what is the next possible date?

The next possible date (that is, one where I can attend the meeting) is 
in May in Xian, China, when the SC32 subcommittee will meet. That will 
include Working Group 2 (WG2). This is not a highly technical meeting 
however. It is also the case that WG2 needs to be educated about Common 
Logic and/or SCL; this process will take more than one meeting.

>
>> This will be a chance for the US part of the international group
>
> What international group? Right now, SCL is being composed by a group 
> which I convened a little less than a year ago and which consists of a 
> small group of people all known to me. The Common Logic group is 
> chaired by Michael Uschold and I believe have not met or had any 
> correspondence for a year now.

The international group I was referring to is ISO/IEC JTC1/SC32/WG2 
which voted to make Common Logic a new work item.
I have just finished my first meeting with them. There were 
representatives from seven countries there: USA, UK, Canada, Australia, 
Korea, China and Japan. This is the group that will eventually approve 
Common Logic as an ISO standard.

If there are actually two logic standardization projects going on, this 
will complicate matters for both me and ISO. We can either try to 
include Uschold's group, or effectively ignore them.

>
>>  to discuss the standard and get their comments out in the open.
>>
>> A Working Draft becomes a Committee Draft (CD) which must be approved 
>> by WG2 and then formally blessed by SC32. I would like for this to 
>> take place at the next SC32 meeting which will take place in Xian, 
>> China in late May 2004. Depending on the outcome of that meeting, we 
>> may need to address comments and prepare another committee draft 
>> (CD2), which will also be subject to SC32 approval. Once CD's are 
>> refined, they become a Final Committee Draft (FCD) with responses to 
>> ballots by SC32. After that it becomes a Final Draft International 
>> Standard (FDIS) which is balloted by JTC1 (the highest level). The 
>> last step is to actually approve the standard as a completely 
>> approved international standard (IS) ISO/IEC IS 24707. My point is to 
>> remind everyone that this a time-consuming process, which will 
>> probably take two or three years. Therefore we have no time to waste.
>
> Well, we have already wasted approximately 5 years, so three more isnt 
> going to make much difference :-)

ISO will cancel the project if we do not show progress within two years.

>
>>
>> The working group likes
>
> ? What working group?? Where was this 'like' expressed, and to whom?

See above. WG2 is ISO's official body to consider drafts of the 
standard and move it through its draft stages to an ISO standard. Each 
country gets one vote, but the group prefers to operate by consensus of 
course. One consequence of seeking standardization is that standards 
are reviewed by many people who are going to vote on its acceptance. I 
believe WG2 has a legitimate process for evaluating the technical merit 
of Common Logic.

>
>> the general organization of having the Common Logic core (now called 
>> SCL) as part 1 of the standard, and then having KIF, CGIF, and CLML 
>> as parts 2, 3 and 4 showing how all of the core can be mapped to each 
>> particular language L. Parts 2, 3, and 4 should each describe 
>> extensions to the languages that are NOT part of the core as well. 
>> There can be informative (i.e., not prescriptive) annexes that help 
>> further explain the languages.
>
> This is not exactly the organization that I favor, and in fact it does 
> not really make sense, since the concrete SCL syntaxes (KIF, CG etc) 
> are not extensions to SCL.  There are two orthogonal issues: 
> concretizing the abstract syntax in various ways, and 
> restricting/extending the language in various ways. Particular 
> concrete syntaxes may not support all language extensions.

I am quite open to any of various ways of organizing the standard. 
Particular attention should be paid to conformance issues which as most 
of you know means how to specify that a particular language or system 
either does or doesn't conform. Levels of conformance are certainly 
acceptable, but if there are too many levels, or too many defined 
extensions/restrictions, then the standard will be quite weak and 
implementers are unlikely to bother with it.

>
>> I might also mention that calling Common Logic a "draft international 
>> standard" is incorrect -- it is currently only an ISO work item with 
>> no content, since there is no official status to the white paper that 
>> was used for voting on the work item. After the SC32 meeting in May 
>> we may actually call it a draft standard. Protocol, terminology and 
>> procedure are very important to the ISO.
>
> Thanks for the correction; I understand that protocol is important, It 
> should perhaps be noted that SCL will be offered to other 
> standardization bodies in addition to ISO.

I would be interested to know which bodies those are. ISO is generally 
the most respected of the international standardization bodies -- that 
is why it takes lots of effort to get an ISO standard approved. If SCL 
is to be offered elsewhere, I can see a whole host of problems, not the 
least of which is having more than one international group wanting to 
change the same document.

>
>>
>> ACTION ITEMS:
>> - prepare a working paper (Working Draft is a specific ISO term) for 
>> the INCITS L8 meeting in mid December. This is the U.S. 
>> representative to ISO's Working Group. As editor, I will take the 
>> lead in keeping an "official" version.
>
> What exactly do you see your role as being? Our problem is that the 
> technical writers of the spec are all running on empty as far as spare 
> time is concerned, and the sheer effort of writing is likely to be our 
> main barrier. I tend to compose in HTML, so if you can transcribe this 
> into Word that would be a great help.

In order to be considered by ISO, we'll have to use their Word 
templates, which are non-trivial and a pain in the butt. That is simply 
a fact we have to deal with. I am happy to keep the "master" or 
"official" version in Word, but it will greatly slow down the process 
for me to have to "transcribe" everyone's changes into word. This is 
especially true considering the special symbols we are using.

Once we decide to go down the ISO standardization road, the process is 
owned by ISO, not us. As editor, I am ISO's manager (so to speak) of 
the project -- I gather together opinions, prepare the documents, 
respond to detailed critiques by the countries' standardization 
organizations (including the U.S.!). The working group WG2 is 
responsible for the overall quality and consistency of the standard as 
it goes through the stages; as editor, I act as liaison between the WG2 
and technical experts. It is therefore almost essential that we have 
experts from the other countries involved in the process.

>
>> Be aware that we will have to use Microsoft Word format -- either 
>> Word itself or Open Office, etc.
>> - attempt to get additional participation from other countries (e.g., 
>> France, Australia, Germany) in commenting on the working paper.
>> - examine other standards that may overlap CL, especially the Z 
>> standard (ISO/IEC 13568) which has been recently approved. There is 
>> concern that Z (pronounced "zed") has already standardized a form of 
>> logic. I have a copy of that standard for anyone who is interested.
>
> We have examined that, at John's suggestion. It has very little to do 
> with SCL, either in its aims (Z is aimed at proving properties of 
> programs, not ontology development) or its syntax (Z syntax is highly 
> idiosyncratic) or its semantics (Z is not a first-order language),and 
> I have seen no evidence of Z being used in any of the likely use 
> groups or user communities for the SCL standard (database technology, 
> description logics, ontology development, semantic web, AI-KR).  I 
> have heard the "concern" you mention expressed only by John Sowa: 
> until I hear evidence of its being a general concern, I am not 
> inclined to pursue the matter in any more detail. The issue was not, 
> apparently, raised at the presentation given in Santa Fe in January.
>
>> As editor, my role is to keep things moving and be custodian of the 
>> "official" version. I intend to take this role seriously, which means 
>> that I have an obligation to try to resolve differences and lead/push 
>> us to agreement.
>
> Please try to do that through me, as chairman and convener of the 
> group. We have had a rather fractious history and it would  be very 
> easy to be diverted into revisiting old debates.

Pat, I have a great deal of respect for you and what you are doing. As 
long as going through you doesn't delay the process too much, that is 
fine with me. But see my comments above -- eventually there are going 
to be lots of people reviewing and commenting on the paper. This will 
involve compromises, explanation and seemingly unnecessary steps. That 
is how standards work, as I'm sure you already have found from your W3C 
work.
>
>> I consider it vital to reach consensus
>
> It is vital that we reach agreement, not necessarily consensus 
> agreement. To achieve consensus in this group may well be impossible 
> and is almost certainly NP-hard.

This is my biggest worry right now. Standards are the ultimate "writing 
by committee" and not for the faint-hearted. We need to develop a 
standard that will work and that people will use. Holding out for one's 
favorite feature is exactly what has stalled the standard and which 
many people still believe will eventually kill it.

Harry