Toward Ontology-Based
Component Composition
Kenneth Baclawski and Artan Simeqi
College of Computer Science
Northeastern University
Component Composition
 Increasingly common software development
technique.
 “Feature interaction” is a serious problem.
 Validation is very hard and rarely done.
 Prospects for improvement are dismal:

Lamport predicted that it would not be a
practical concern within the next 10 years.
Pseudo-Programming Languages
 General purpose languages for specifying
component composition.
 Only support limited kinds of component
connector and system requirement.
 The main technique for specifying
component composition today.
 Actually make it harder to validate
compositions! (Lamport 1997)
Back to Basics: Mathematics
 Mathematical proofs are already inherently
compositional.
 Mathematics is more general and more
powerful than any PPL.
 Why are we trying to improve on
mathematical reasoning?
Ontology
 Shared understanding within a community.
 Formal, declarative semantic model
 An ontology may include:




Vocabulary terms
Taxonomic and other relationships
Constraints, inference rules and assertions
Theorems, lemmas and corollaries
 An ontology is a mathematical theory.
Ontology-Based Computing
PPL
Domains
 Apply general mathematical reasoning to
specialized domains specified by ontologies.
 By contrast, PPLs apply specialized
reasoning without any domain specificity.
Ontology
Mathematical Constructs
Suggested Approach
 Theory is the fundamental building block.



Theories can be composed using the colimit.
A theory can be interpreted using concrete models
Ontologies, requirements, environments and
component specifications are theories.
 Justification (proof trace) is a formal proof
represented as a data structure that can be
queried and manipulated using automated
tools.
Metamodel for Ontology-based
Component Composition
Proof Trace
1..*
verifies
comb ines
Theory
Composition
1
1
Ontology
Requirement
produces
Component Specification
Environment
Dynamic Composition and
Reconfiguration for Sensor Systems
 Extreme test case for ontology-based
component composition
 These are hard real-time systems that can
operate at a nanosecond time scale.
 Validation for a reconfiguration must occur
without human interaction in seconds while
a mission is occurring.
 Lives depend on correctness.
Conclusion
 Ontology-based computing is emerging as
an important new computing paradigm.
 Basing component composition on
ontologies has many advantages:




Grounded in general mathematical reasoning
Tractability can be addressed
Close connection with the domain language
Uses the compositionality of mathematics.