Institute for Information Systems IWi Institut (IWi) für at the German Research Wirtschaftsinformatik Center for im DFKI Saarbrücken Artificial Intelligence (DFKI), Saarland University Semantic EPC: Enhancing Process Modeling Using Ontologies Dr. Oliver Thomas, Michael Fellmann
Agenda 1. Motivation 2. Framework for the semantic annotation of business process models 3. Ontology development 4. Use cases, benefit 5. Conclusion 2
Semi-formal business process models Characteristics Company A Company B Easy to model Pretty close to natural language 3
Problems of semi-formal business process models Company A Company B Ambiguity, lack of clearly defined semantics Not machine interpretable, therefore hard to query, compare, validate and transform other than manually 4
Semantic annotation of business process models Framework Conception Representation Ontologies OWL Metadata RDF Models XML 5
Semantic annotation of business process models Framework Conception Representation Ontologies OWL Metadata RDF Models XML 6
Towards a core Ontology for sbpm Resources Ontologies Domain specific Enterprise Ontology TOVE BMO General purpose (upper) ontologies CyC SUMO WordNet BPM languages and -methods BPMN ARIS/EPC UML Activity Diagram Problems of ontology merging Contradicting definitions Highly interdependent definitions Different methods such as top-down, bottom-up or middle-out result in different conceptualizations Heterogeneous ontology languages Lack of authority, relevance Merging of all relevant ontologies will not be beneficial Filtering and selection of concepts is needed 7
How to find relevant classes and instances? BPM languages and existing models as a source Complex Process order receipt Event Function Atomic Receipt of order Sales Europe Process customer order Customer order Org. Unit Org. Unit Europe Sales Europe XOR Customer oder Customer document Document accepted rejected Database Inform. carrier Classes can be extracted from languages, instances from models 8
How to find relevant classes and instances? BPM languages and existing models as a source Complex Process order receipt Event Function Atomic Function Sales Europe Receipt of order Process customer order Customer order Events Org. Unit Org. Unit Europe Sales Europe XOR Customer oder Customer document Document Organisation accepted rejected Data Database Inform. carrier Classes can be extracted from languages, instances from models 9
How to find relevant views (Categories) Analysis of methods and frameworks 10
Selected views (top-level Categories) for the SBPM-Ontology Views 11
SBPM-Ontology Structure semantic-business.org 12
Semantic annotation of business process models Framework Conception Representation Ontologies OWL Metadata RDF Models XML 13
Overall Approach uses assignedto Task partof Event Service dependson Org. unit related Task Sales event Production event Data service Verification service uses Rule processing dependson dependson uses Production assignedto Sales assignedto partof partof verification partof partof receipt uses acceptance uses Customer data uses Product Data Prodct Configurator Creditworthiness confirmation Creditworthiness rule verification rule Feasibility check Creditworthiness check semtype semtype semtype semtype semtype semtype flow flow Event 1 flow Gate Gate Function 1 XOR 1 Event 2 Event 3 Function 2 Function 3 XOR 2 Event 4 flow flow flow flow flow flow Process- Model received Verify order accepted rejected Send order confirmation Send order rejection processed Connections: Mapping EPC-Model/Ontology Instantiation Specialisation Property EPC control-flow 14
Benefits Querying with Inference Which function depends on which rule? SPARQL-Query PREFIX s: <http://semantic-business.org/2007/02/sepc#> SELECT?func?rule WHERE {?func rdf:type s:function.?func s:semtype?t.?t s:haspart?p.?p s:uses?rule.?rule rdf:type s:rule } Rule uses verification partof haspart Creditworthiness check uses uses Creditworthiness Creditworthiness rule semtype Function 1 Verify order Result?func?rule --------------+----------------------- s:verify s:creditworthinessrule Connections: Mapping EPC-Model/Ontology Instantiation New facts Property that are not Inferred contained Property in the original model can be inferred 16
Benefits Further examples Validation using Rules (e.g. SWRL) Integration of rules in the ontology for advanced reasoning and validation Example: A car rental company wants to introduce the following rule: Each process model in which the function rent sportscar occurs, there must be a subsequent function 'check age > 25. Validation on the semantic level Semantic standardization of models Consistent usage of terminology provides better understanding spanning enterprises Reduced effort of transformation in executable models Enhanced process modeling environments are possible Display of context information using reasoning, e.g. products or organizational units Suggestions on how to avoid semantic errors Potential benefits regarding querying, validation, modeling and transformation 17
Some obstacles and Problems Increased modeling effort Knowledge and skills regarding semantics Selection of instances for annotation during model creation Complexity of the ontology must be hidden Dynamics, changes of the ontology Who is responsible for the process modeling ontology? How can ontology evolution be handled? Ontology evolution must be community-driven 18
and a possible solution: Semantic collaborative BPM Aspects Scenario Model process using a standard tool and language. Import of the process model in a wiki-like environment. Process-Annotation as a collaborative effort. Collaborative tools may solve some of the problems 19
Conclusion Our contribution Framework for the semantic annotation of business process models Core ontology for business process annotation Future research directions IT-support for the semantic annotation of business process models Use cases 20
Thank you for your attention! Questions? 21