1. A Practical Ontology-Driven Workflow Composition Framework Huy Pham, Deborah Stacey, Rozita Dara School of Computer Science University of Guelph Guelph, Ontario, Canada

2. Slide 2 of 11 Quick Overview A brief survey of ontology-driven approaches to workflow composition (ODWC) Proposal: A more modular and reusable approach to planning-based ODWC Knowledge Engineering and Ontology Development 2011

3. Slide 3 of 11 Intro and Motivation Automated workflow composition A great tool to help non- expert users to overcome the expertise gap The task of finding a sequence of actions that accomplishes a given goal (i.e., planning) Knowledge Engineering and Ontology Development 2011

4. Slide 4 of 11 Intro and Motivation Real-world WF problems are often knowledge-intensive, and hence can benefit from an ontology-driven approach Standardized semantics Expressive Reasoning services Problem: Many existing approaches either don't use planning, or do it in less reusable and modular ways Knowledge Engineering and Ontology Development 2011

5. Slide 5 of 11 Existing Approaches Interactive Composition E.g., Hlomani, et. al. WFs are composed interactively using inputs from user Provide assistance instead of design proposals Template-based Composition E.g., Morik, et. al. WF designs are suggested from a pre-built library of successful WF built by experts Cannot help in unseen cases Planning-based E.g., Bernstein, et. al. WF designs are proposed by planning algorithms Adhoc, less reusable Knowledge Engineering and Ontology Development 2011

6. Slide 6 of 11 How About? Potential Benefits: Loose coupling --> Reduced complexity + Increased reusability Reusable compositional knowledge Knowledge Engineering and Ontology Development 2011

7. Slide 7 of 11 A case study An intelligent student advisor: Helps university students select courses, taking into account: Core requirements Course prerequisites Student objectives Requirements for Course Selection knowledge Reusable  Course selection knowledge must be modeled in an ontology Modular  Kept separated from other knowledge Rich & Effective  Capture and use of expert advices Knowledge Engineering and Ontology Development 2011

8. Slide 8 of 11 Course Ontology Knowledge Engineering and Ontology Development 2011

9. Slide 9 of 11 Course Objective Ontology Knowledge Engineering and Ontology Development 2011

10. Slide 10 of 11 Planning Ontology Knowledge Engineering and Ontology Development 2011

11. Slide 11 of 11 Discussion What worked: Course selection What didn't: More elegant way of soliciting user's objectives More planning constraints More details in: Our other paper, "Practical Goal-based Reasoning in Ontology-Driven Applications" Our website, http://ontology.socs.uoguelph.ca/ Knowledge Engineering and Ontology Development 2011