A Model of Events for Integrating Event-based Information in Complex Socio-technical Information Spaces
Download as PPTX, PDF0 likes996 views
(1) The document presents a formal ontology model called Event-Model-F for integrating event-based information across complex socio-technical systems. (2) Event-Model-F is based on the foundational ontology DOLCE+DnS Ultralight and defines events using a pattern-oriented approach and six core ontology patterns. (3) The goal of Event-Model-F is to provide a common understanding and representation of events to allow different event-based systems to efficiently communicate and share information.
A Model of Events for Integrating Event-based Information in Complex Socio-technical Information Spaces
- 1. A Model of Events for Integrating Event-Based Information in Complex Socio-technical Information Systems Ansgar Scherp, Thomas Franz, Carsten Saathoff, Steffen Staab Institute WeST University of Koblenz Germany http://west.uni-koblenz.de/
- 2. Emergency Response Scenario Calls to report about a power outage Fire Department Coordinate and Emergency Citizen keep up to Documentary Report Hotline support date and update • Several emergency response entities are involved about the incident Creates incident with audio recording • Using different event-based systems Reports Emergency Report and update • Common understanding of exchanged multimediaincident by taking photos Control Center about the information is needed to efficiently communicate etc. Coordinate and keep up between ER entities Request to to date Police Department Forward report about a Liaison flooded cellar Emergency Response Officer Coordination EU Integrated Project WeKnowIt http://www.weknowit.eu/ Snapped pole image from: http://www.dailymail.co.uk/ Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 2
- 3. Outlook Emergency Response Scenario Motivation Formal Model of Events Existing Event Models Future Work Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 3
- 4. Motivation Events need to be modeled and are useful in a variety of application domains Lifelogs, multimedia experience sharing Emergency response Cultural heritage News Sports Surveillance … However Event detection and annotation from different sources Using different data models and proprietary solutions Event descriptions need to be shared between systems Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 4
- 5. Event-Model-F • Humans like to think in terms of events & entities • Human-centered approach to capture experience and knowledge • Events • Occurrences in which humans participate • Subject to interpretation and discussion • Development of core ontology Event-Model-F • Sophisticated modeling support for occurrences in which humans participate • Homage to event model E by Westermann & Jain Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 5
- 6. Requirements to a Common Event Model • Participative aspect • Temporal aspect • Spatial aspect • Structural aspect • Mereology (composition) • Causality • Correlation • Interpretation • Experiential aspect (documentation) Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 6
- 7. Comparison to Existing Event Models SsVM = Semantic-syntactic video model VERL = Video event representation language CIDOC CRM = Conceptual reference model for cultural heritage Web Science and Ansgar Scherp Event-Model-F 7 Technologies scherp@uni-koblenz.de Slide 7
- 8. Ontology Patterns of Event-Model-F • Event-Model-F defines six core ontology patterns based on Description and Situation pattern (1) Participation pattern (2) Mereology pattern (composition) (3) Causality pattern (4) Correlation pattern (5) Documentation pattern (6) Interpretation pattern • Specified in Web Ontology Language (OWL) • Formalized in Description Logics • Graphical representation in UML-like notation Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 8
- 9. Modeling Basis of Event-Model-F • Ontology pattern Descriptions and Situations (DnS) as fundamental design principle for Event-Model-F • Formal representation of context through use of roles • Decoupling concrete events and objects from their roles in a specific contextual situation • Description • Specification of roles required in a specific situation • Can be understood as template • Situation • Observable real-world situation, i.e., a concrete combination of events and objects • Satisfies a description, if it fits into the template Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 9
- 10. Example: Descriptions and Situations Pattern • DomesticPowerOutage- • DomesticPowerOutage- Description defines roles Situation defines objects • AffectedBuildingRole House-1 : Object • AffectedPersonRole Paul-1 : Object, • … Classify Sandy-1 : Object,… … • Important: Different people may claim different causes for the outage • Different interpretations of the same DomesticPowerOutageSituation a) Snapped power pole b) Problem with the power plant Image source: Wikipedia Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 10
- 11. (3) Causality Pattern • Event (cause) implies other event (effect) • Causal relationship holds under some justification • Causes and effects are events, and only events EventCausalityDescription Concept defines defines exactly 1 Cause EventCausalityDescription Role defines exactly 1 Effect EventRole defines exactly 1 Justification Description defines only (Cause or Effect Effect Cause or Justification) Justification satisfies isSatisfiedBy exactly 1 EventCausalitySituation classifies isRoleOf Situation Event Description Example: The event of a snapped power pole causes a isEventIncludedIn isObjectIncludedIn power outage. EventCausalitySituation Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 11
- 12. (1) Participation Pattern • Participation of living and non-living objects in events • Reuse of domain knowledge EventParticipationDescription defines exactly 1 DescribedEvent defines min 1 Participant defines some LocationParameter defines some TimeParameter Roles the defines only (DescribedEvent or Participant or entities play LocationParameter or TimeParameter) isSatisfiedBy exactly 1 EventParticipationSituation Real world entities Example: Firemen and home owner are involved in an incident of a house fire. Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 12
- 13. (2) Mereology Pattern • Composite event consists of multiple component events • Composition along time, space, and space-time defines EventCompositionDescription EventCompositionDescription Concept Parameter defines exactly 1 Composite Description EventRole defines min 1 Component isParameterFor EventCompositionConstraint defines only (Composite or Component or EventCompositionConstraint) Composite Component TemporalConstraint classifies classifies parametrizes isSatisfiedBy exactly 1 EventCompositionSituation SpatioTemporalConstraint satisfies Time-Interval parametrizes Event hasParticipant Spatio-Temporal-Region isEvent Example: Events of a snapped power pole, power IncludedInObject SpatialConstraint outage, and bursting ofhasQuality are components ofparametrizes Situation hasQuality a dam a larger flooding event.Quality hasRegion isTime Space-Region isSpaceTime isSpace EventCompositionSituation IncludedIn IncludedIn IncludedIn Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 13
- 14. (4) Correlation Pattern • Correlate events have a common cause • Happen at the same time or share some overlap • Useful, as often only correlation is observable and the common cause remains unknown defines EventCorrelationDescription EventCorrelationDescription defines min 2 Correlate Concept defines exactly 1 Justification Description EventRole Role defines only (Correlate or Justification) satisfies isSatisfiedBy exactly 1 EventCorrelationSituation Correlate Justification classifies classifies Situation Event Description Example: Several correlating power outage events isEventIncludedIn happen in the city. EventCorrelationSituation isObjectIncludedIn Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 14
- 15. (5) Documentation Pattern • Provide documentary evidence for an event • Annotation of events with photos, video, audio, etc. EventDocumentationDescription defines exactly 1 DocumentedEvent defines some Documenter defines only (DocumentedEvent or Documenter) isSatisfiedBy exactly 1 EventDocumentation- Situation Example: • Documenter classifies ImageData defined in COMM (Core Ontology on Multimedia) • Formal model of MPEG-7 low-level descriptors Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 15
- 16. (6) Interpretation Pattern • Explicit modeling of contextual views on events • Combines the instantiations of patterns (1) to (5) EventInterpretationDescription defines EventInterpretationDescription defines exactly 1 Interpretant Role defines min 1 RelevantSituation RelevantSituation Description defines only (Interpretant or RelevantSituation) Domain Ontology isSatisfiedBy exactly 1 EventInterpretationSituation EventRole RelevantComposition RelevantCausality satisfies Interpretant RelevantCorrelation For example: Interpretation of a power outage classifies RelevantParticipation • Citizen: power outage on our street is caused by snapped classifies power pole Situation • Officer: power outage of the city is caused by a problem Event Situation isEventIncludedIn in the power plant EventInterpretationSituation isObjectIncludedIn Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 16
- 17. Design Approach 1. Chose of foundational ontology DOLCE+DnS Ultralight as modeling basis • Aims at capturing the most essential aspects in the world • Defines disjunctive upper classes Event, Object, Quality and Abstract • Follows a pattern-oriented approach for ontology design 2. Use of ontology design patterns • Generic solution to recurring modeling problem • Reduces complexity of the designed model 3. Defining Event-Model-F as core ontology • Provides structural knowledge that spans across multiple domains, e.g., lifelogs, emergency response, etc. • Build on top and align it with DOLCE+DnS Ultralight Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 17
- 18. Comparison to Existing Event Models • Event Model E, EventML, Event Calculus, CIDOC CRM, VERL, SsVM, Event Ontology, Eventory • Do not follow such a systematic development approach • Semantically ambiguous • Conceptually narrow • Hinders interoperability of different event-based systems Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 18
- 19. SemaPlorer Place Object Type Event Web Science and Ansgar Scherp Event-Model-F 19 Technologies scherp@uni-koblenz.de Slide 19
- 20. Future work on Event-Model-F • Extraction of events and objects from Web content • Reasoning on Event-Model-F with Linked Geo Data • Event-Model-F Website • Provides the ontology and examples in OWL • Implementation of Java API • http://west.uni-koblenz.de/eventmodel/ Web Science and Ansgar Scherp Event-Model-F 20 Technologies scherp@uni-koblenz.de Slide 20
- 21. Thank you for your attention! Questions? Ansgar Scherp scherp@uni-koblenz.de http://west.uni-koblenz.de/ Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 21
- 22. --- Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 22
- 23. What is an event? • Events • Perduring entities that unfold over time • Occurrences in which humans participate • Subject to discussions and interpretations by humans • Objects • Enduring entities that unfold over space • Events and objects require each other Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 23
- 24. Ontology Stack • Domain Ontologies • Cover a specific domain • Example: fishery, human body, emergency response, etc. • Core Ontologies • Coverage: span across multiple domains • Examples: annotation, communication, events, ... • Foundational Ontologies • Span across multiple core ontologies Domain Ontologies Core Ontologies Foundational Ontologies Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 24
- 25. Non-functional Requirements • Extensibility • Include future aspects for describing events • Axiomatization & formal precision • Required for a common understanding of events • Interoperability between systems • Modularity • Reduce complexity by selecting only what is required • Reusability • Share common events/objects for different interpretations • Reuse of domain knowledge • Separation of concerns • Core model needs to be applicable in many different domains • Separate structural knowledge from domain-specific knowledge Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 25
- 26. Non-functional Requirement • Extensibility • Pattern-oriented approach of DOLCE+DnS Ultralight • Specializing/extending existing patterns, adding new patterns, … • Axiomatization & formal precision • Foundational ontology DOLCE+DnS Ultralight as basis • Semantically precise through Description Logics • Modularity • Pattern-oriented design • Reusability • Integrating existing domain ontologies • Separation of concerns • Structural knowledge is defined in the ontology design patterns • Domain-specific knowledge is linked through classifying roles Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 26
- 27. Event-Model-F API • Programming interface to Event-Model-F • Enable direct use of the Event-Model-F without requiring to know the internal details of the ontology • Layered architecture of the API Your Application Event-Model-F Extended API Event-Model-F Core API RDF Storage (Sesame) • Release under open source license https://launchpad.net/eventmodelf Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 27
- 28. Short Example: Serious Weather Conditions • During serious weather conditions a flood happens • Causality: power pole snappes and causes a power outage • Participation: citizen observes this event from his home Web Science and Ansgar Scherp Event-Model-F Technologies scherp@uni-koblenz.de Slide 28