ISO 15926 as OWL - An Ontological Approach to Data Warehousing using OWL: ISO 15926 as OWL - An Ontological Approach to Data Warehousing using OWL David Price, Tore Christiansen, Matthew West, John Kendall 8th NASA/ESA PDE Workshop April 2006


Agenda: Agenda What is ISO 15926? Why OWL? An example related to Ship dataset integration


Courtesy Matthew West (Shell): Courtesy Matthew West (Shell)


ISO 15926: ISO 15926 Strictly speaking ISO 15926 is Integration of life-cycle data for process plants including oil and gas production facilities In reality it is a model for a completely generic data warehouse you first add your domain model to the warehouse as “Reference Data” you then add “domain model instances” as data too a 4D standard upper ontology 4D = 3 space dimensions plus 1 time dimension the core of an approach for doing deep semantic analysis of a domain


Integration Architecture: Foundation Concepts General Concepts Discipline-specific Concepts Derived concepts Integration model Integration Architecture


Standardization: Standardization ISO 16926-1 Overview and ISO 15926-2 Data model became International Standards 2003 Part 2 is written using the ISO EXPRESS modelling language Also known in the Oil and Gas industry as the “EPISTLE Core Model” Much of the modelling was through the POSC-CAESAR consortium Currently in-process in ISO ISO 15926-4 Reference Data ISO 15926-3 Geometry and topology are in-process ISO 15926-7 Implementation methods for data exchange and integration is also in-process


Philosophy of ISO 15926 (West): Philosophy of ISO 15926 (West) Metaphysics - The study of the nature of things Likely to be relevant to Integrating data models and data sets from different sources Reference: Metaphysics - an Anthology, Blackwell, ISBN: 0-631-20279 Most of our everyday view of the world is based on notions that are more than 1000 years old cf most of our everyday view of physics is still dominated by Newton. There has actually been some progress, particularly in the last 200 years, that can be of considerable value in modelling the world around us. Quine, Russell, Lewis, Armstrong, Wittgenstein, Frege, Cantor, Peirce and many more.


3D Ontologies (West): 3D Ontologies (West) A 3D ontology treats physical objects as 3D objects, or continuants, that pass through time 3D objects are wholly present at each point in time Physical objects are viewed from the present The default is that statements are true now Physical objects do not have temporal parts Different physical objects may coincide The object-at-a-point-in-time is the object of primary interest To talk about an object at different times it is necessary to time index statements in some way, e.g. X at t. A 3D ontology also has 4D objects in it


4D Ontologies (West): 4D Ontologies (West) A 4D ontology treats all individuals, things that exist in space-time, as 4D objects or spatio-temporal extents Things in the past and future exist as well as things in the present 4D extent is viewed from outside time rather than in the present Individuals have both temporal parts and spatial parts Two individuals with same spatio-temporal extent are same thing The object over its whole life is the object of primary interest 4D object is not (usually) wholly present at a point in time, but its whole is extended in space as well as time The object at a point in time is a temporal part of the whole Change is naturally expressed through a 4D classical mereology


Individual/state and temporal whole-part: Individual/state and temporal whole-part


The data model: The data model The next few slides show bits of the ISO 15926-2 EXPRESS model Just to give a flavor of what it’s about There are around 200 entity types in the entire schema Note that many of the concepts are also found in the meta-models of languages like OWL and UML


Ontologies and OWL: Ontologies and OWL Given the ontological nature of ISO 15926, it should be no surprise that EXPRESS is not quite sufficient It’s all that was available in the mid- and late-1990s At one time the ISO 15926 team worked on, but never completed, a logic language for the model Things have moved on in the past few years The Web Ontology Language (OWL) became a W3C Recommendation in 2004 ISO Common Logic standard is in-process OMG is working on the Ontology Definition Metamodel which includes OWL, Topic Maps, Common Logic, etc.


Recent 15926 team activity: Recent 15926 team activity Translate the ISO 15926-2 Core Data Model concepts into OWL mapping directly into the OWL language itself where appropriate For example, 15926 EXPRESS entity “relationship” represented as OWL Object Property not an OWL class named “relationship” “relationship” is actually one of the current open issues, Is it an OWL Class or OWL ObjectProperty?


Technical details: Technical details The representation tries to exploit the capabilities of OWL While keeping useful characteristics of the current Express representation Entity types are represented as OWL classes, properties and individuals Specialization is represented by sub-classing between OWL classes Classification is represented by instantiation between OWL classes Classes can be defined in terms of restrictions Relationships become properties with domain and range Property characteristics include functionality, symmetry and transitivity


Resulting ontology: Resulting ontology The resulting representation is OWL full And is therefore not guaranteed to be computationally determinant but there is no guarantee that reasoners will not work


Information pyramid (Christiansen): Information pyramid (Christiansen)


An example (Christiansen): An example (Christiansen)


One usage scenario: One usage scenario STEP AP EXPRESS 15926 OWL Upper Ontology 15926 OWL Reference Data OWL Individuals STEP AP OWL Subclass of Type of Reason over STEP data integrated in 15926/OWL data warehouse configurable P21 data


Class of individual OWL: Class of individual OWL


Class of relationship in OWL: Class of relationship in OWL


Ship AP Integration Project (Kendall): Ship AP Integration Project (Kendall) Process scope handover of data from shipbuilder/design agent/prime contractor to MoD definition of the as-built condition Configuration control and feedback to design processes e.g. for refit Main information scope ship common model, design, manufacturing also, Ship general particulars, principal dimensions, hull framing specification, plating, openings, edge cut-outs, plating arrangements, welds, profiles, longitudinal framing arrangements, Hold, tank and compartment definitions


Test data: Test data Used SEASPRITE test data ‘364P_fil.stp’ ‘Eleo Maersk’ Tanker construction block From Odense Steel shipyard Based on ARM model Semantics contained within the ARM model Mapping from ARM to AIM defined in the AP document


Assumptions made: Assumptions made The STEP models are unclear as to whether the entities represent, classes of things, designs of things or actual things. For example: ENTITY Ship has the capability to be all three. This means it maps to Class, Functional Physical Object or Materialized Physical Object For the purposes of this project, it was assumed that the entities referred to designs of objects. i.e. Functional Physical Object Use the P21 ID as the OWL identifier


Specialization example: structural_system class possible_ individual ID136 Specialization example ID136 is an arranged_individual ID136 is a bulkhead bulkhead is a type of structural_system class_of_functional_object is a type of class class in an entity type possible_individual is an entity type class of functional object bulkhead arranged_ individual a specialization relationship, the circle indicates the subclass


Ship classes as OWL: Ship classes as OWL


Ship High Level API: Ship High Level API Specify a generic API for accessing Ship Product Data Documented in UML (Magic Draw UML) for a Java implementation Generic API realized by access to an OWL data store


AP218 High Level API: AP218 High Level API


T1.3 Demonstrator Architecture: T1.3 Demonstrator Architecture


Conclusions: Conclusions ISO 15926 provides a semantic framework for classifying and representing ‘things’ It provides a consistency check for data models Upper ontology approach shows promise Ship project highlighted a number of inconsistencies and poor modelling in the ISO 10303 ship APs. confusion over classes of thing, designs of things, actual things attributes can be classification, properties, identifications ISO 15926/OWL Issues It seems that there is not one right way to map data models and data to the ISO 15926 model (part 2) Feedback to ISO 15926 team = add XML Schema datatypes ISO 15926 was EXPRESS so only had EXPRESS datatypes