WebOnt Use Cases

Dear all: her comes a summary of our use case considerations.
Regards

Ruediger

_________________________________________________________________________

DaimlerChrysler Use Cases
=========================

The following use cases are primarily directed towards engineering aspects of 
semantic web applications. Besides those there are lots of business or other 
aspects which have been dealt with by other people - so we do not repeat them 
here.

0. Use Case

We would like to provide support for engineering developers. In our use case, 
an engineering developer sits in front of her CAD system, executing some 
engineering task. She has at the same time access to engineering best practices 
documented in a semantic web context. Using the underlying ontology, we can 
filter the information, to provide just those aspects relevant to her. Typical 
questions that arise, are to filter information based on context, to provide a 
view on the ontology adapted to a certain task (e.g. classification may provide 
a different view than catalog browsing), and finally constraints between the 
elements can be used to compute values for certain properties.

More precisely, we are modeling a very complex world, with many objects, 
properties and relations. These models should be built by many different 
people, more or less concurrently. I.e. a team of people is developing the 
model of our world, not a single person. As many people are modeling the same 
world independently, we would like to provide a guiding structure to the 
modeling people involved. A central task is to ensure the consistency between 
the various modifications/edits executed by the people involved. For example, 
in a group meeting, we may agree that in our world we should model engineering 
parts, production process steps and materials in a certain way. These are 
somehow related, in some agreed fashion. This commonly agreed framework then 
serves to guide the actual modeling. The people that execute the modeling can 
now specialize engineering parts, production process steps and materials, 
define properties of the specializations, and place them in one of the agreed 
upon relations.

To summarize:
(a) ensure consistent modeling of the ontology by several people;
(b) varying views on the ontology based on user context (taxonomy versus 
classification when selecting screws);
(c) constraints to express relations between elements.

Some of the aspects important in this context can be generalized in the 
following way:

1.Engineering information

Engineering information is mainly characterized by the following points:

- large bodies of knowledge: catalogs of materials, parts and components; bills 
of material; supplier information; geometric, simulation and other models...
- a great variety of interrelationships between different parts of engineering 
information: requirements, functional descriptions, behaviours, structures, 
geometry, ...
- arithmetics and especially geometric models are integrated parts of 
engineering information.

A whole universe of engineering IT tools exists today each especially dedicated 
to a certain aspect of engineering problem solving: from large database 
systems, PDM (product data management)  tools, CAD, numerical simulation 
systems, work flow, etc.

Today, greater parts of this complex information are not at all represented in 
the IT systems, or they are left implicit inside procedural code.

Engineering is essentially a collaborative effort: many people interact in 
engineering problem solving - contributing their special views on the task to 
be solved.

As a consequence, the many different IT systems used today in typical 
engineering environments are 
- restricted in their interoperation capabilities (using standards like STEP), 
- need special "hand made" interfaces for interoperation, or 
- can only exchange data where the semantics is hidden.

2.  Semantic Web in Engineering

From a Semantic Web point of view, we have three main aspects of dealing with 
engineering information:

2.1.)
large, semantically rich engineering ontologies dealing with requirements, 
functional descriptions, behaviours, structures, geometry, arithmetics, etc.;
2.2.)
different views on engineering information - reflecting the views of the 
various people involved in collaboration (or similarly, integrating different 
ontologies in a consistent way);
2.3.)
changing (but related) content: versions, (parallel) variants, succeeding 
stages of models, work flow, etc.

3. Semantic Web Requirements

We do not expect that the Semantic Web as currently under discussion will 
really solve these problems. They are too deep and too widespread. But in order 
to be usable as a platform for an improved treatment of engineering problems 
the Semantic Web should provide some basic capabilities:

3.1.) structuring large ontolgies/bodies of information:
- hierarchical and modular structures of ontolgies are supported
- meta knowledge about ontological knowledge (including tagging and reification)

3.2.) different views
- user specific views can be defined as part or on top of ontologies
- different ontologies can be merged (partially)

3.3.) arithmetics and geometry
Beyond typical description logics engineering ontologies need rich 
arithmetics/geometrical modeling capabilities.

4. Practical Requirements

In order to make that work three pre-conditions should be fulfilled:

4.1.)
general purpose and application-independent but domain-specific "standard" 
ontologies (about functions, structures, behavious, geometry, etc.) should be 
provided as a "starting point" for application specific information models;
4.2.) 
sophisticated tools are needed in order to deal with complex enginerring 
ontologies: for retrieval and for consistent knowledge modeling (also be 
experienced end users - esp. for the "lower" and more frequently changing parts 
of the ontologies).
4.3.)
sophisticated techniques are need for database and tool integration: the huge 
amounts of engineering information available today in "non-SemanticWeb form" 
must be usable.

Received on Monday, 3 December 2001 13:47:28 UTC