Some Industrial Experiences in the Development and Use of Ontologies

Some Industrial Experiences in the
Development and Use of Ontologies
Prof Matthew West
Shell Information Technology
International
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Agenda
• Experiences and Learnings in Shell
• Initial Paradigm, Principles and Framework
• An example of applying the principles and
framework
• From Shell to International Standards
• An example of differing interpretations of a
standard data model
• Evolution of the Paradigm
• Individuals as spatio-temporal extents
• Classes, physical properties, and set theory
• Conclusions
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Experiences and Learnings in Shell
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Differing data models for the same thing in different
parts of Shell c1990
Comp.
Canada
C.I.A.
SUKO
MF
Aus.
Model
BOSS
CMF
Arch.
Eng.
Arch.
MF
EP
SNR
SNC
Matrix
HydroC. DataMdl Ref.Mfg. Data Str. MCSM
Area
Org'n
Loc'n
Product
Equip.
Facility
Contract
Plan
Purch.
Prod'n
Stock
Sale
Price
Account
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Main causes of problems found in data models
• Constraints in the data structure – sometimes
deliberate, sometimes inadvertent – prevented data
from being held.
• Only a current snapshot was held, change and
history could not be managed.
• Classes that were essentially the same or overlapped
were not recognized as such. So customers and
suppliers would be managed separately, without
recognising that a customer could also be a
supplier.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Context and scope
Scope
Context
• The scope of a model must fit within a defined
context for internal consistency.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Context and scope
Outside context
• Models which are developed in different contexts
will be incompatible.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Context and scope
Conceptual data models developed in the same context will fit together.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Initial Paradigm, Principles and Framework
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Principles for High Quality Data Models
1. Entity types should represent, and be named after, the underlying
nature of an object, not the role it plays in a particular context.
2. Entity types should be part of a subtype/supertype hierarchy (class
hierarchy) in order to define a universal context for the model.
3. Activities and associations should be represented by entity types (not
relationships or attributes).
4. Relationships (in the entity/relationship sense) should only be used
to express the involvement of entity types with activities or
associations.
5. Entity types should have a local identifier within a database or
exchange file. These should be artificial and managed to be unique.
6. Candidate attributes should be suspected of representing
relationships to other entity types.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
THE GENERIC ENTITY FRAMEWORK
PERSON
ORGANIZATION
FACILITY
NATURAL UTILITY
CREDIT/DEBIT ACCOUNT
PROPERTY
Flashpoint
Production Facility
Weight
Calendar
Density
Department
Storage Facility
Transport Facility
Business
Size
1
Temperature
2
MATERIAL
3
4
5
6
A registered
LPG cylinder
Pack
# AV78
A Case of goods
A human being. What makes someone an
individual, i.e. the spirit rather than the body.
A group of people organized for a purpose,
e.g. company, department, government, club,
NGO.
The service to which one or more
pieces of equipment may be put.
Includes: CD4, P10A;
Excludes: Pump No. 1224 serving as P10A.
The service a natural (as opposed to manmade) material provides, e.g. an oil reservoir.
A credit/debit account holds the state of
indebtedness between two parties, e.g. the
French Railways account with Shell France for
diesel fuel.
ACCOUNT ITEM
Account Item and Account Class Template
ASSORTMENT
INFORMATION CONTENT
MONEY
Batch #2245 in Tank
Truck #57
A characteristic that is the physical or chemical
nature of something, e.g. its temperature.
TIME
Shell
Services
International
MATERIAL
Account Class
Account Item
Account
Code
Includes: Packed Product, Equipment, Batch
Description
of
Composition
(of Account
Item)
Time Work Done
10345
General Platform Maintenance
4
10506
Water filtration renewal project
10
10432
P.E. training course
December
Pa c k # AV 78
TIME SHEET - FEBRUARY
A thing that consists of matter and has mass.
using
PROJECT PLAN
Account Item
has
Materials
used in
Remove seals on
pump X
Facilities
Classification
Planning
of
Classification
(of Account
Item)
6
20
Classification
(of Account
An account item represents the official view of
life, e.g. a line on a timesheet.
Class)
PURPOSE OF THE GENERIC
FRAMEWORK
Schedule
People
Subject of
Schedule
of
as
Subject of
Account Class
used in
An information content that represents the
indebtedness between two parties, e.g. the
information on bank notes.
The meaning that can be derived from how
material is arranged, e.g. the content of a
document for which you can claim copyright.
An assortment is a set of unlike things, e.g. a
project plan consists of materials, facilities, people
& timescales.
A period or point in time on the time continuum,
e.g. 20:52:02 on 22nd December 1993.
SPACE
Entity
MEASUREMENT
Subject
ACTIVITY
ASSOCIATION
TOKEN
Transfer
Composition
Life
Measurement Template
Measurement
of
Physical
Transfer
Material
measured by
for
measured in
Characteristic
Type
reported in
Unit of
used to report
following
followed for
using
used to perform
Qualifiers
A portion of the continuous expanse in which
things exist and move.
A measurement is an estimate of a
characteristic of something derived by
comparing the characteristic against some
standard, e.g. taking the temperature of a liquid.
Measure
Measurement
Method
Cost = distance X rate per mile
Time
Period
Version
Facility
Point in Time
Succession
Natural Utility
Storage
Ownership
Measurement
DEEM
How much salary?
Deem
Transfer Template
Request
Actual: Something that exists or has existed.
Expected: An activity or object for which it is expected
an actual activity or object will exist in the future.
FACT & FICTION
Space
3. Activities and associations should be represented
by entity types (not relationships).
Fact: A department has five people.
Expected
Surface
Account Item
1. Entity types should represent, and be named
after, the underlying nature of an object, not the role
it plays in a particular context.
2. Entity types should be part of a subtype
/supertype hierarchy in order to define a universal
context for the model.
Life Cycle
Fiction: What if one person is removed ?
Required
5. Candidate attributes should be suspected of
representing relationships to other entity types.
Fact: What is.
Fiction: What is not.
Planned
4. Relationships (in the entity/relationship sense)
should only be used to express the involvement of
entity types with activities or associations.
Point
Predicted
Installation
A value that is decided (rather than estimated),
e.g. the deeming of my salary following annual
review.
ACTUAL & EXPECTED
Class
Line
Control
Calculation
A class is a type of something for which a basis
for inclusion or exclusion is defined.
An instance is an occurrence of something.
Actual
Credit/Debit
Account
Estimate
KEY PRINCIPLES
TESCO
A123 JKL
Essence
Determine Value
The determination of a value from other values
using some derivation method, e.g. the cost of
transporting fuel from refinery to gas station.
J667 DEF
Typical
Organization
Shell
Specific
Volvo Motorway Master
Classification
Transfer of
Control
CALCULATION
SAFEWAY
Class of Truck
CHARACTERISTIC
Property
Collection
Transfer of
ownership
Measuring
Facility
INSTANCE/CLASS
Instance
Person
Assembly
Instantiation
MATERIAL
OVERVIEW
The generic entity framework has been developed to
provide a basis for producing consistent data
models. Without a consistent basis, models
produced in isolation will vary arbitrarily and be
difficult to reconcile. On the other hand,
independently developed model with the same basis
will tend to be consistent.
The generic entity framework is so called because it
is a subtype/supertype hierarchy of generic entity
types.
6. Entity types should have a single attribute as their
primary unique identifier. This should be artificial
and not changeable by the user.
ASSIGNMENT
Assortment
Employment
Facilities
Transferable
PHYSICAL
Agree
Money
TRANSFER
People
From
TRANSFER
OF
OWNERSHIP
Source
Organisation
Employment and Assignment Template
Money
Authorize
Person
To
Destination
Bank Account
An association that indicates that a person
works in a particular position, e.g. Fred Bloggs
has been assigned to the head of IT.
Fact
Involvement
Location
Facility
TRANSFER
OF
CONTROL
Reality
Information
Content
Assignment
Materials
TRANSFER
of
Fiction (what if)
Employment
employee in
Person
(of Person)
of
Assignment
(of Person)
as
assignee
in
filled
through
Organization
by
employer in
Composition
(of
Organization)
Position
of
has
from
in
PHYSICAL TRANSFER
TRANSFER OF OWNERSHIP
TRANSFER OF CONTROL
AGREE
AUTHORIZE
REQUEST
of
JO'S
Transfer
TRANSPO RT
Customer
FOR SALE
EMPLOYMENT
Supplier
from
to
B718 JRX
The physical transfer of some product from one
place to another, e.g. transfer of material from
the warehouse to the loading bay.
Physical Transfer Template
The transfer of ownership of something from
one person or organization to another, e.g.
ownership of the house is transferred from
Alison White to Joe Cotter.
Transfer of Ownership Template
of
Material
transferable
property
(of Material)
of
created by
creates
to
Facility
destination
from
source
Business
Location
Concluding a negotiation to the satisfaction of
the parties involved.
Asking for something, e.g. customer requesting
service from a supplier.
Material
of
controllable
Composition Template
COMPOSITION
Transfer (of
creates
Ownership)
terminated by
to
terminates
from
destination
Organization
Transfer
of
(of
Entity
Serial No. 5755/A
Classifies
A123 JKL
source
controller
by
An association that indicates one thing is
assembled as part of another, e.g. a motor
assembled into a truck.
Collection
An association that indicates one thing is a
member of a set of things (not necessarily
identical), e.g. a fleet of tankers.
VERSION
CONTROL
Restricted
Composition, Version and Succession Template
SUCCESSION
An association that indicates a piece of equipment
is acting as a facility, e.g. installation of Bloggs
100H pump serial no. 5755/A as P10.
An association that indicates something is of a
particular class, e.g. the classification of a particular
truck by model Volvo Motorway Master.
12 July, 2016
Classification Template
Installation Template
of
caused by
Entity
Composition
John has control
of the fork lift.
from
ORACLE
RDBMS
ORACLE
RDBMS FOR
UNIX
Barrel
An association that indicates that something
belongs to a person or organization, e.g. the
ownership of the house by Frank Gunning.
An association that indicates something is
being looked after by a person or organization,
for their use or under their supervision.
versionable
Installation
MICROSOFT
EXCEL
VERSION 5.0
MICROSOFT
EXCEL
VERSION 4.1
classifiable
of
classifiable
Entity
version
Succeeded by
Class
successor
‹#›
 SHELL SERVICES INTERNATIONAL LTD.
© Shell International Oil Products Ltd
by
Facility
performer
Organisation
An association that indicates one version of
something succeeds another, e.g. Excel 5.0
succeeds Excel 4.1.
Slide
of
classifier
by
as
duty
predecessor
by
Classify
The copyright of this document is vested in Shell
International Petroleum Company Limited, London.
All rights reserved. Neither the whole nor any part of
this document may be reproduced, stored in any
retrieval system or transmitted in any form or by any
means (electronic, mechanical, reprographic,
recording or otherwise) without the prior consent of
the copyright owner.
installable
classifier
of
Succession
An association that indicates something is a
version of a higher level object.
Equipment
of
(of Equipment)
by
ORACLE
RDBMS FOR
VAX VMS
causes
Classification
composite
component
of
Version
An association that indicates something is
contained within something else, e.g. oil stored
at a depot, oil stored in a barrel.
Volvo Motorway Master
TESCO
P10
destination Organization
Filing cabinet
Depot
SAFEWAY
J667 DEF
Shell
© Shell Information Technology International Ltd
OWNERSHIP
DU02
CD1
Shell
used in
Assembly
from
terminates
SIPC
Bloggs 100H Pump
has
using
to
by
employment
Qualifiers differentiate between different types of a
subject. Qualifiers consist of sets of mutually
exclusive concepts that can apply to a subject. They
are still generic entity types thus any object belongs
to one and only one from a set, ever. Conversely,
any object does belong to one of the subtypes from
each set of qualifiers.
CLASSIFICATION
INSTALLATION
(COLLECTION)
of
controllable
Control)
terminated by
source
owner
by
STORAGE
COMPOSITION
Material
Composition
created by
Alison White
An association that indicates a person works
for a person or organization, usually in return
for direct remuneration.
Transfer of Control Template
Control (of
Material)
property
of
Officially endorsing something for a purpose,
e.g. authorizing a plan for execution.
(ASSEMBLY)
of
Ownership
Physical
Transfer
The transfer of control is the act of one person or
organization handing over control of something to
another, e.g. control of this truck is transferred to
Mr. Keith Logan.
SUBJECTS & QUALIFIERS
Subjects are things that are described by the same
types of characteristic. Thus the subject material is
described in terms of mass, volume, etc. It does not
matter whether you are interested in an actual
material (something you can touch), a planned
material (something you can't touch) or a material
class. It is properties like mass and volume that are
relevant.
by
terminated by
terminates
An example of applying the principles and framework
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Example: Personnel & Security
Personnel
Telephone_directory
Security
• personnel_no
• name
• address
• sex
• start_date
• nat_ins_no
• company
• department
• date_to_department
• salary
• name
• reference
• room
• tel_no
• badge_no
• name
• room
• company_department
• date_issued
• date_released
• badge_type
• Entity types from three different systems
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Resolution of the Telephone Directory entity type
allocated S[0:?]
in S[0:?]
Telephone
allocated S[0:?]
Person
Room
works_in S[0:?]
Organization
Department
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Partial resolution of the Personnel entity type
Person
lives_at
S[0:?]
works_in S[0:?]
Organization
Address
Department
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Company
Slide
© Shell International Oil Products Ltd
‹#›
Partial resolution of the Security entity type
allocated S[0:?]
Person
Room
issued_with S[0:?]
Security_
badge
works_in S[0:?]
Organization
Department
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Company
Slide
© Shell International Oil Products Ltd
‹#›
Entity type duplication/complex entity type
resolution
allocated S[0:?]
in S[0:?]
Telephone
allocated S[0:?]
Person
Room
issued_with S[0:?]
lives_at
S[0:?]
Security_
badge
works_in S[0:?]
Organization
Address
Department
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Company
Slide
© Shell International Oil Products Ltd
‹#›
From Shell to International Standards
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Engineering Data Handover
Engineering Contractor
Design
Owner/Operator
Fabrication
Drawing Office
Historically paper
Maintenance
Engineering
Data
Warehouse
Engineering
Data
Warehouse
Operations
© Shell Information Technology International Ltd
Commissioning
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
The World of STEP for the Process Industries
PIEBASE
EPISTLE
POSC/Caesar
PISTEP SPI-NL
PIPPIN
POSC
PlantSTEP Activity ProcessBase
PDXI
Model
EPISTLE
Core Data Model
Epicentre
ENAA
ISO 15926 Life-Cycle Integration
AP221 - Functional and Schematic data
AP227 - Plant Spatial Configuration
© Shell Information Technology
International
Restricted
12 July, 2016
Slide
AP231
-LtdProcess Definition
SC4 Data Integration Architecture
© Shell International Oil Products Ltd
ISO TC184/SC4 - Industrial Data
‹#›
An example of differing interpretations of a standard
data model
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Facility and some key relationships
physical thing
product
transformation
basis
installable
containable
installation
service
© Shell Information Technology International Ltd
containment
facility
Restricted
container
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Facility Definition
A facility is a functional thing and is the capability to perform a function.
A facility is a service to be provided, duty to be performed, functional view
of, placeholder for, or a logical view of something.
For example, tag number P1102A is a placeholder for, say, a boiler
feedwater pump. If the particular pump that is fulfilling that function
becomes defective, we can swap out the specific pump that is installed
and replace it by another. The tag number does not change.
A facility will often be implemented by something man-made, but it could
equally be implemented by a natural physical_thing (a natural utility).
Examples of a facility are :–
–
–
–
–
P10, the pump service at the bottom of a column
21T103, the measuring facility at the top of a column
The TAMANA Platform TEP-A
C: drive on your PC
The Shuaiba formation of the oil reservoirs in the Fahud field in Northern
Oman
– The Atlantic Ocean (taken, for example, as a transportation medium)
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
INSTALLATION
DU02
CD1
Bloggs 100H Pump
P10
Serial No. 5755/A
An association that indicates a piece of equipment
is acting as a facility, e.g. installation of Bloggs
100H pump serial no. 5755/A as P10.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Evolution of the Paradigm
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Two dominant paradigms
• A continuant/occurrent based foundation,
where continuants endure through time, but
are wholly present at each point in time
when they exist, and occurrents that exist in
time as well as space.
• A four-dimensional approach that sees all
objects as spatio-temporal extents, and
physical objects extended in time as well as
space.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
The Decision
• The former paradigm was found to be counter-intuitive to the approach
we already had of all things having a lifetime. The latter paradigm,
introduced to us by Partridge, on examination:
– Mapped easily with the approach we already had to managing change.
– Had the possibility of a clear and unambiguous identity basis based on
spatio-temporal extent.
– Reflected better the intuitions in the existing data model.
– Provided a rigorous approach for analysis.
– A classical mereology, which is both strong and simple is sufficient even
when dealing with changing objects.
• The EPISTLE Data Modelling Team decided to rework the EPISTLE
Core Model using this 4D paradigm. The result can be found in ISO
15926-2:2003. After the fact, on comparing our results with the work of
Sider, we found that our models were consistent with his preferred
paradigm.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Individuals as spatio-temporal extents
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Metaphysics, Ontology and Integration
Space time map
A particular
spatio temporal
extent
3D
SPACE
TIME
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
State/Sub-state
Time
period
3D
SPACE
Sub state
state
TIME
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Materialised Physical Object
event 1
event 2
3D
space
B
A
D
C
Time
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
thing
possible_
individual
relationship
part
composition_of_
individual
whole
temporal_whole_part
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Replaceable Parts
3D
SPACE
installed
removed
installed
removed
US President
Clinton
Bush
© Shell Information Technology International Ltd
Restricted
TIME
12 July, 2016
Slide
Masolo et al “Figures could be considered as some kind of mereological fusion of the
player-stages of a given role …”
© Shell International Oil Products Ltd
‹#›
Classes, physical properties, and set theory
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Some approaches to Set Theory
X
Y
M
N
O
A
1
Z
B
2
© Shell Information Technology International Ltd
3
P
C
Restricted
4
5
12 July, 2016
6
Slide
© Shell International Oil Products Ltd
‹#›
thing
1
abstract_
object
possible_
individual
classified
class
classifier
relationship
classification
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
class_of_class
class_of_individual
colour
possible_individual
blue
red
my car
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
abstract_object
relationship
superclass
class
© Shell Information Technology International Ltd
specialization
subclass
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›
Conclusions
• Although the ISO 15926-2:2003 data model
is a foundation ontology, and is highly
principled in its content and structure, this
has not been a matter of fiat, but of
evolving experience, trying out ideas to see if
they work, and adopting new ideas when
they prove their worth in practice.
© Shell Information Technology International Ltd
Restricted
12 July, 2016
Slide
© Shell International Oil Products Ltd
‹#›