Re: REQDOC: NEED INPUT! - [Content Interoperability]

Jeff and all,

Here was the information from our Content Interoperability F2F breakout
session that Jonathan Dale scribed. The first is our final use case and
requirements list, the second is the actual session details (both in
text). We have already communicated internally about your needs based on
today's telecon's discussion and will be trying to revise the 2 use
cases (AgentCities-Travel Planning and Web Portal/OntoWeb). 

As I suggested at the telecon, you might want to take the base
requirement descriptions from our final F2F document at:
http://lists.w3.org/Archives/Public/www-webont-wg/2002Jan/0066.html

Best,
Leo

Jeff Heflin wrote:
> 
> Hello,
> 
> On Jan. 17, I asked for the chairs of the use case groups to provide
> input for the Requirements Document. I have still not received any. Of
> course, my original request said by "Wed. Jan 24," and since Wednesday
> was actually the 23rd this may have led to some confusion. Hopefully, no
> one interpreted it as meaning "some later year where Jan. 24 falls on a
> Wednesday." ;-)
> 
> Anyway, I'd like to reiterate my request. If we have to, we'll simply
> use the read-ahead documents for the face-to-face as our input, but then
> we risk losing valuable information and insights that came out of the
> breakout sessions at the face-to-face. So please, try to get us this
> input by the end of the day. If you have nothing to add to the original
> use case documents, then let us know and we'll begin from there. Thank
> you for your cooperation.
> 
> Jeff Heflin
> 
> p.s. I've attached the original message below...
> 
> --------------------------------------------------------------------------
> 
> Hello everyone,
> 
> It was a pleasure meeting many of you in New Jersey this week. We are
> preparing to edit the requirements document and need your input. If you
> were the chair of one of the breakout sessions on use cases please send
> us information on the following items produced by your session:
> 
> Detailed representative use cases (about one page each)
> Detailed design goals, if any (about one page each)
> Requirements (short paragraphs, also mention which use case or design
> goal motivated the requirements)
> 
> Please collect this in ASCII text format and post it to the mailing list
> with a REQDOC: prefix by Wed. Jan 24. Of course, you are welcome to
> solicit input from members of the original use case subgroup, but we
> would prefer if each group only posted a single message. If you are
> unable to meet this timeframe, please appoint someone from your use case
> group to perform this task. If my memory serves me correctly, the chairs
> of the use case breakout sessions were:
> 
> Content Interoperablity - Mike Dean or Leo Obrst
> Collection Management - Guus Schreiber
> Web Services - Stefan Decker
> General Requirements - Deborah McGuinness
> 
> To remind you of the use cases, design goals, and requirements, I've
> attached Jim's original summary of the product of the four breakout
> sessions. Note that I've sorted the requirements by the initial grade
> given by playing "Dan's game." Please send the paragraph for every
> requirement regardless of the grade assigned. We plan to provide things
> that some people see as requirements but were rejected by the group in a
> separate section, and at any rate having them there allows for us to
> still discuss and consider them.
> 
> Thank you!
> 
> Sincerely,
> Jeff, Jonathan, and Raphael
> 
> ---------------------------------------------------------------------
> 
> USE CASES:
> - web site mgt
> - Homogenous collection
> - doc about an object/artifact/design
> 
> - Travel planning
> - Portal from multiple sources
> 
> - Ubiq. Computing (small devices)
> 
> DESIGN GOALS:
> - Shared Ontologies
> - Ontology Extension
> - Ontology evolution
> - Detect Inconsistency
> - Ontology Interoperability
> - Scalability
> - Ease of Use
> - XML syntax
> - Expressiveness
> - Internationalization
> 
> REQUIREMENTS:
> A Annotation/tagging of ontologies (some particular properties)
> A Ontology namespaces/inter-ontology reference
> A ability to state uniq. names
> A character set support
> A lexical representation (internationalization)
> A ontology management language features (versioning)
> A unambiguous term referencing using URIs
> A uniqueness of unicode strings
> B Define range contraints on data types
> B Ontology mapping relations (equivalento)
> B ability to state closed worlds
> B commitment to  ontologies
> B ontology partitioning
> B solution to "tagging/grouping" problem
> B- Class as instance
> B- Relational Types
> B- records (complex datatypes)
> C capability (chaining of properties, transitivity)
> C effective decision procedure
> C layered approach
> C- commitment to portions of ontologies
> X ability to integrate signatures
> X bit efficient encoding
> X defaults
> X multicultural mechanism (view)
> X support for speech acts
> X unique name assumption
> - (procedural attachment)
> - Definitional constraints of conjunctive type
> - arithmetic primitives
> - pre and post conditions
> - support for expressing work flow
> - support for variables
> 
> OPEN ISSUES:
> - defaults appear to be needed, but difficult or impossible
> - presentation syntax
> - explainability

-- 
_____________________________________________
Dr. Leo Obrst		The MITRE Corporation
mailto:lobrst@mitre.org Intelligent Information Management/Exploitation
Voice: 703-883-6770	7515 Colshire Drive, M/S W640
Fax: 703-883-1379       McLean, VA 22102-7508, USA

WebONT Working Group - Content Interoperability Breakout Group

Leo Obrst, James Barnett, Mike Dean, Rafael Volz, Jonathan Dale (scribe)

15th January, 2002



Two Use Cases

1. Agentcities and Travel Planning for small devices

2. Ontology and Content Interoperability with OntoWeb



Language Requirements

1.      Ontology namespaces and inter- and intra-ontology references

2.      Ontology mapping relations (rules)

3.      Consistency annotations; definition of annotation tags

4.      Ontology partitioning

5.      Meta-knowledge

6.      Property typing

7.      Ontology layering (more of a goal)

8.      Lexical representations and muti-lingual/multi-cultural support

Jonathan Dale, PhD
Member of Research Staff
Network Agent Research Group
Fujitsu Laboratories of America
595 Lawrence Expressway
Sunnyvale, California 94085, USA
Tel: +1 408 530 4543
Fax: +1 408 530 4515

WebONT Working Group – Content Interoperability Breakout Group
Leo Obrst, James Barnette, Mike Dean, Rafael Volz, Jonathan Dale (scribe)
15th January, 2001

Use Cases
LO: Requirement is to distill from the list of 22 use cases, 2 specific exemplars. 
And we also need to come up with a list of concrete list of language requirements 
that we can vote on.

LO: Focus should be on content interoperability, and we seemed to focus on 
device interoperability. Main classes are (1) device interoperability, (2) 
organizational interoperability, (3) conceptual search/eCommerce interoperability, 
(4) knowledge fusion/agent interoperability.

LO: Suggest that we focus on knowledge fusion and agent interoperability areas, 
for example, FIPA/Agentcities use cases, such as agents using different 
ontologies. 

JB: It would be good to mention/use knowledge fusion as a use case. Eventually 
all of this will have to be tested, so they may implement their own 

RV: Suggest having a management briefing/executive overview of the whole 
area and has pointers to the more expanded examples in the appendix.

LO: JH will eventually pull together all of the information from each of the 
Breakout groups into one big document. 

RV: Might be better to have the summary, requirements and then the 2 use 
cases.

LO: Three issues drawn from the use cases; (1) mapping between ontologies 
including composition/decomposition of ontologies, (2) different views of 
ontologies based upon physical/functional property matching, and, (3) granularity 
of ontologies, perhaps for small devices.

RV: First cases can be regarded as ontologies mediating between different 
instance data (databases, etc.), i.e. an integrated schema to prevent multiple 
mappings between individual instance.

MD: Third case can be used for deriving (graph pruning) ontologies for smaller 
devices, and, providing a finer granularity view of ontologies. 

RV: We don't want everyone having to implement small device support.

JD: FIPA Gateways Technical Committee use gateways at the edge of networks 
not only to tokenise the data stream to make it smaller, but also content 
pruning/refinement so that the actual data is made smaller for the end device. 
Additionally, the use of gateways means that everyone doesn't have to 
implement this (the gateway owner provides a mapping).

MD: Or, we can fall back on the ISO OSI model by pushing it up into the 
application layer (layer 7).

MD: Can we combine device and organizational interoperability work into one 
use case, since they are related.

LO: Perhaps also bind the agent interoperability into this, too.

JD: <<Insert description of the Agentcities Evening Organizer here.>>

LO: This sounds like your use case, MD; perhaps we can combine this with the 
Agentcities EO to provide an entire stack of issues from devices at the bottom to 
agent-oriented applications at the top.

MD: Aggregation was voted a not a key issue, but we might want to consider it 
here as an organizational issue.

LO: Perhaps we should expand the specific use case for Agentcities (B) to a 
more specific case along the lines of the Agentcities EO (devices, agents, 
brokers, services and different ontologies for each).

RV: We will need an example of content interoperability, for example OntoWeb; 
online ontology collaboration and content/annotation collaboration.

LO: This falls under the organizational interoperability general case (23).

MD: Internationalization also falls under the Agentcities/device use case.

JD: FIPA has been thinking about translation between ontology terms in different 
languages via the FIPA Ontology Service, which is implemented using OKBC 
and KIF.

LO: Combine use cases 1 (Travel Planning) and 2 (Agentcities), so that we have 
two exemplars of Agentcities and OntoWeb.

Language Requirements
LO: General categories: (1) Correlation, mapping, merging and translating of 
ontologies (inter-ontology), (2) device/resource interoperability, (3) ontology 
creation and management, and, (4) views, context and language support.

LO: (1) Add separate a requirement of how to make references between different 
ontology representation languages (for example, OWL to UML, OWL to KIF). 
This is a prime requirement for FIPA, since it will not advocate a single ontology 
representation language.

RV: Mapping between two different logics is difficult, since it requires second-
order logic which makes it undecidable.

LO: Add a separate requirement for aggregation.

Section 3.1
Keep:
1. Include inter- and intra-
3. Add relations
12. Rename 'Consistency annotation'; ability to annotate portions of the ontology, 
defining the annotation tags in the language.

Reject:
2., 6., 8., 11. Part of 3.
4., 5. Part of the Web Services group
7., 10. Out of scope, not in the ontology representation language
9. Not a language feature

Section 3.2
Keep:
5. 

Reject:
1., 2., 4. Out of scope
3. Covered by Web Services
6., 7. Out of scope, part of the device specific issues
8., 9., 10., 11. Out of scope, runtime issues

Section 3.3
Keep:
3.
4.
9. Not a requirement, but a goal, and it is important to reiterate

Reject:
1., 2. Covered by Web Services
5. Covered by XMLS data types
6. Covered by 9.
7. 
8. Out of scope, language goal
10., 11. Covered by Web Services

Section 3.4
Keep:
5. Add in text of 7.

Reject:
1. Too descriptive
2. Rules are required for this, or a separate ontology view/presentation language
3.
4. Covered by 3.1 (1)
6. Covered by 5.
8.
9.