1. Programmierung verteilter Systeme Lab Institut für Informatik Universität Augsburg Universitätsstraße 14, 86159 Augsburg Tel.: (+49) 821/598-2174, Fax: -2175 URL: http://www.informatik.uni-augsburg.de/vs Semantic-Enabled Software Engineering Position Paper Bernhard Bauer

2. © Bernhard Bauer, all rights reserved 2006 Semantic-Enabled Software Engineering Motivation + Introduction Problems of Todays SE Application Areas An Idea of a Semantic-Enabled Software Engineering IDE

3. © Bernhard Bauer, all rights reserved 2006 Motivation Economic Trends  Mergers and Split of Enterprises  Outsourcing / Offshoring  Industrialisation of SW Development Technology Trends  Model-Driven Development  Generation and Transformation  Semantic Web Technologies  Necessary  Interoperability  Semantic-based  Integration of heterogenuous Systems  Automation of SW Development Software Engineering + Semantic Technologies

4. © Bernhard Bauer, all rights reserved 2006 Model-Driven Software Development Model Driven Development: Computational Independent Model (CIM)  describes the business (logic)  defines business processes and domain specifics Platform-Independent Model (PIM)  describes a software system that supports some business  independent from any implementation technology Platform-Specific Model (PSM)  developed/generated from the PIM  depends on the underlying technology Target platform implementation  mappings to multiple middleware platforms Agent Model Java/EJB Model BPEL Model PIM Agent Impl Java/EJB Impl BPEL Impl Other Impl PSM CIM An old-fashioned figure

5. © Bernhard Bauer, all rights reserved 2006 Semantic Technolologies You know Them!

6. © Bernhard Bauer, all rights reserved 2006 Problems of Today‘s Software Development Methodologies:  Methodologies (e.g. RUP, Agile Development Methodologies, V-model XT) customized to enterprises  Guidelines und document templates not supported in the necessary detail Interoperability issues:  mainly at run-time  mergers arise problems  build-time model-driven integration of systems neglected. Search and composition of components / models:  Re-use of COTS components and services difficult, on code-level  manually lookups based on syntactical descriptions  lack of semantic rich descriptions of e.g. functionality and quality attributes  semi-automatic orchestration of components to value-added services difficult. Syntactic-based model transformations:  Model transformations on the basis of meta-models, no semantics  reuse of model transformations and adjusting model transformations to new versions of meta-models.

7. © Bernhard Bauer, all rights reserved 2006 Possible Solution Mainstream Software Engineering (UML + MDSD) + Semantic Technologies (Ontologies, Reasoning, etc.) = Semantic-Enabled Software Engineering

8. © Bernhard Bauer, all rights reserved 2006 Ideas for Semantic-Enabled Software Engineering UML + Semantics – not really innovative Requirements Engineering / Knowledge Acquisition  Develop Domain Ontologies  Re-use this ontologies through the whole development process Use Case Diagram  Define your roles in an ontology  Prepare use cases with IOPE Activity Diagram / Interaction Overview Diagram  Model your process with IOPE  Define static and dynamic (!) semantics of process Class Diagram  Add additional ontological concepts to class diagrams  …  Starting point for further exploiration of Semantic Technologies

9. © Bernhard Bauer, all rights reserved 2006 Ideas for Semantic-Enabled Software Engineering Semantic-Enabled Model Transformation MM2MM1 Reference Ontology Bootstrap Model Transformation Binding (sem. Annotation) Model Transformations Metamodels v1.5 v1.6 Metamodels v2.0 v2.1 Binding (sem. Annotation) Inference Component Ontology-based Model Transformation Model Manipulator Sem-MT-Component Source Target input relationship (binding)

10. © Bernhard Bauer, all rights reserved 2006 Ideas for Semantic-Enabled Software Engineering Model Analysis Model Analysis, e.g.  Does the model satisfy / not satisfy specific properties  Formalization of the semantics wrt. previous models in the development process  „verification“ of the system  Is the composed model consistent  Is the model correct wrt. the modeling guidelines

11. © Bernhard Bauer, all rights reserved 2006 Ideas for Semantic-Enabled Software Engineering Model Synthesis Model Synthesis  Integration of different models  from different development teams  Different enterprises  In new models like Interaction Overview Diagrams  Like service synthesis on the Semantic Web Service level for  Components  Models  Processes  Data  Etc.

12. © Bernhard Bauer, all rights reserved 2006 Ideas for Semantic-Enabled Software Engineering Attribution of Meta-Models Modelling Guidelines are pre-requisite for Model Transformation Annotate the meta-model with additional information  to describe guidelines  allow meta-model optimization  „data flow“ to handle dynamics of models

13. © Bernhard Bauer, all rights reserved 2006 Semantic Enabled IDE MM x MxMx ModelRep.Ontologies ApplOnt x Ref.Ontology Reasoner Model Manipulator Sem-X-Component RuleSet Sem-X-Tools SemIDE-Infrastructure Semantic enabled IDE 1 2 3 4 5...

14. © Bernhard Bauer, all rights reserved 2006 Conclusions Semantic Technologies allow  to add semantics and meaning to models  to analysis and synthesis models  to generate initial model transformations  to automate the software engineering process Solve Shortcomings of Today’s Software Engineering  Methodologies  Interoperability issues  Search and composition of components / models  model transformations