ontology-based context modeling · challenges and problems • open topic in the semantic web...
TRANSCRIPT
![Page 1: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/1.jpg)
Universität InnsbruckLeopold Franzens
making semantics real.3rd Workshop on Context Awareness for Proactive Systems
© Copyright 2006 DERI Innsbruck www.deri.at 19 June 2007
Ontology-Based Context Modeling
Reto Krummenacher, Thomas Strang
![Page 2: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/2.jpg)
making semantics real.2 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Background
Digital Enterprise Research Institute– Applying semantics to Web services to achieve (semi-)
automatic discovery, composition, …
– Non-functional and dynamic aspects of service descriptions and goals/tasks to execute
– Middleware for ‚Internet of Services‘: semantic tuplespaces• Management tasks / Non-functional properties• Self-Representation• Reflection, scalability trade-offs
![Page 3: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/3.jpg)
making semantics real.3 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Overview
1. Ontologies: Some Basic Facts2. Modeling Criteria3. The Survey4. Challenges & Problems
![Page 4: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/4.jpg)
making semantics real.4 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Ontologies Some Basic Facts
Universität InnsbruckLeopold Franzens
making semantics real.©
Copyright 2006 DERI Innsbruck www.deri.at 13 May 2007
![Page 5: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/5.jpg)
making semantics real.5 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Ontologies
Ontologies are„explicit formal specifications of the terms in a domain and the relations among them“
Modeling characteristics– semi-structured with clear model semantics (not OO)– modeling facilities for concepts and properties (not Logic)
Projecting real-life entities onto machine-understandable data constructs
![Page 6: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/6.jpg)
making semantics real.6 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Benefits
• Interoperability– high-level, explicit specification (understanding)– reusability, applicability (speed of implementation)– data and system integration (in the large)
• Validity and compatability checking, formal constraints
• Reasoning– validation of models and instances– derivation of instances & relations (implicit knowledge)
• the system can infer more about the big picture– knowledge interpretation and evaluation
![Page 7: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/7.jpg)
making semantics real.7 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Ontology Languages
• Two branches of languages– First-Order Logic (FOL)
• Description Logics (DL; e.g. OWL-DL) subsets of FOL– Logic Programming (LP)
• FOL, DL– open world, no unique name assumption– subsumption reasoning, consistency checking, classification
• LP– closed world, unique name assumption– query answering, consequence finding (rules systems)
? – child(?x) AND gender(?x,male).
![Page 8: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/8.jpg)
making semantics real.8 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Success factors for context modeling ontologies
Universität InnsbruckLeopold Franzens
making semantics real.©
Copyright 2006 DERI Innsbruck www.deri.at 13 May 2007
![Page 9: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/9.jpg)
making semantics real.9 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Modeling Criteria: Context (1)
• Comparability of data values– heterogeneity of coding systems, units and values
• Traceability– provenance (trust)– computational source for derived context
• Logging, history– decisions based on the past– monitoring (detecting unlikely changes)
![Page 10: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/10.jpg)
making semantics real.10 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Modeling Criteria: Context (2)
• Quality– e.g. mean error, standard deviation
• Satisfiability (constraint modeling)– restrictions and constraints on acceptable values
• Inference, derivation– high-order context (situations, activities…)
• inWater, moving swimming– new contextual types based on primitive values
• show the relationship of speed with distance and time
![Page 11: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/11.jpg)
making semantics real.11 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Modeling Criteria: Ontology (3)
• Reusability– simple and small ontologies (DC, FOAF)– genericity: domain independent (upper ontologies)
• Consistency– no contradictions (neither implicit nor explicit)
• Completeness, redundancy– cover the whole domain– but do not redefine explicit/implicit knowledge
![Page 12: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/12.jpg)
making semantics real.12 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Modeling Criteria: Ontology (4)
• Readability– humans develop ontologies, humans choose ontologies– understandable, intuitive relations and terms– not important for machines, but...– very relevant factor for reuse, and adaptation
• Language, formalism– choose the right language for the problem– choose the right reasoning support for the problem
• compatability of formalism• decidability (FOL + LP!!!)
![Page 13: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/13.jpg)
making semantics real.13 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
The Survey
Universität InnsbruckLeopold Franzens
making semantics real.©
Copyright 2006 DERI Innsbruck www.deri.at 13 May 2007
![Page 14: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/14.jpg)
making semantics real.14 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Ontologies
MobiLife Classification-based situational reasoning for task-oriented mobile service discovery
ConOnto Negotiation Context Information in Context-Aware Systems
SCAFOS / SCALA A first-order logic model for context-awareness in distributed sensor-driven systems
CONON / ULCO Ontology-Based Context Modeling and Reasoning using OWL;
Metamodel for Context A Metamodel Approach to Context Information
CAPNET RDF based Model for Context aware Reasoning in Rich Service Environments
CARE Loosly Coupling Ontological Reasoning wiht an Efficient Middleware for Context-awareness
SOUPA SOUPA: Standard Ontology for Ubiquitous and Pervasive Computing
MAS / mySAM Representing Context in an Agent Architecture for Context-Based Decision Making
CDF Context Description Framework for the Semantic Web
CAMidO CAMidO, A Context-Aware Middleware Based on Ontology Meta-Model
GAIA An infrastructure for context-awareness based on first order logic
OO Quality Model A resource and context model for mobile middleware
VTT Finland Managing Context Information in Mobile Devices
COBRA-ONT An ontology for context-aware pervasive computing environments
CoOL / ASC CoOL: A Context Ontology Language to enable Contextual Interoperability
DOLCE-DnS Understanding the Semantic Web through Descriptions and Situations
CoDAMoS Towards an extensible context ontology for Ambient Intelligence
GAS GAS ontology: an ontology for collaboration among ubiquitous computing devices
CWI-Context Modeling Adaptation in Web Services Execution using Context Ontologies
UbiComp, Pervasive, PerCom (CoMoRea), CONTEXT, CAPS,
PUC, IEEE Pervasive...
![Page 15: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/15.jpg)
making semantics real.15 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Introduction
• Aims to show– the state of the art– deployed features and factors that lack support– examples of work done and work to be done
• Definitively not a complete list of efforts– new examples monthly...
• Difficult to find complete information about models– ontologies not publicly available– lack of complete descriptions
![Page 16: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/16.jpg)
making semantics real.16 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Observations (1)
• Genericity:– abstract vocabularies to describe context values
• ConOnto (ContextView, ContextFeature), • ASC (Aspects, Scales, ContextInformation)
– upper ontologies to model entities involved in context- aware systems: Person, Location, Environment, Application, Device…
• SOUPA, CONON, CoDAMoS (user, service, platform,…)
• Context information is not (only) profiles– profiles with values (formalized) in ontologies– key-value approaches with (formalized) values
![Page 17: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/17.jpg)
making semantics real.17 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Observations (2)
• Comparability– seldom explicitly integrated (values not at the core)– how to compare non-countable values?– counter-example ASC: focus on the values
• Traceability– VTT Finland framework attaches attribute source
• no further modeling of sources– CONON tags values with type of source
• sensed, derived, aggregated, deduced
![Page 18: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/18.jpg)
making semantics real.18 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Observations (3)
• Logging, history– often done by use of timestamps– GAIA: integrates relational database for temporal
queries (values regularly stored in RDBMS)
• Quality– Most clearly recognized meta-context
• probability, confidence, meanError• baysien reasoning, fuzzy logic• quality ontologies
![Page 19: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/19.jpg)
making semantics real.19 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Observations (4)
• Satisfiability– logical expressions (rules)– external services (application and Web service bindings)– again: of particular interest for non-countable values
• Derivation, inference– integration of derivation rules, axiomatic expressions
• activity(sleeping) <- location(inBed) AND eyes(closed)– inter/intra operations of ASC
• Speed = Interoperation(Distance, Time)
![Page 20: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/20.jpg)
making semantics real.20 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Observations (5)
• Most models rely on FOL (in fact OWL-DL)– subsumption reasoning, entity hierarchies, model checking
• LP is chosen for inclusion of context rulesforall X suggestion(X,drink) <-
X:human[activity->running] and T:temperature[value->V, unit->Celsius] and V > 20 .
• Few combined solutions for schema and value modeling (FOL) and the integration of derivation and user rules (LP)
• Interesting: CDF extension to RDF– trueInContext, contextProbability properties
![Page 21: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/21.jpg)
making semantics real.21 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Survey: Conclusions
• Quality and timestamps recognized as important
• Provenance needs more attention– yet more important in large-scale distributed settings– also a prerequisite for trust measures
• Interoperability crucial– solid ontology modeling– comparability, constraints modeling (satisfiability)– especially in open pervasive environments and the
“Internet of Services”
![Page 22: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/22.jpg)
making semantics real.22 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Challenges & Problems
Universität InnsbruckLeopold Franzens
making semantics real.©
Copyright 2006 DERI Innsbruck www.deri.at 13 May 2007
![Page 23: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/23.jpg)
making semantics real.23 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Challenges and Problems
• Top-down creation– the applications determine the models– resulting ontology on a per case basis. Reusability?
• Applications become globally reachable– not some tiny tool on a mobile device– need for standardization, or integration, mapping
• Accessibility– Reuse of ontologies requires that they are available– Lack of publicly available ontologies: a human-caused
problem
![Page 24: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/24.jpg)
making semantics real.24 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
Challenges and Problems
• Open topic in the Semantic Web community– combination of FOL and LP
• causes undecidability, ongoing research– Description Logic Programs– SWRL (OWL + RuleML), RIF (W3C WG)– WSML-Full
– scalability, performance of reasoners– distributed querying and reasoning
• Future: “Internet of Services”?– context-aware discovery, composition, negotiation– combination of functional and non-functional aspects
![Page 25: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/25.jpg)
Universität InnsbruckLeopold Franzens
making semantics real.3rd Workshop on Context Awareness for Proactive Systems
© Copyright 2006 DERI Innsbruck www.deri.at 19 June 2007
Thank you.
Reto Krummenacher Digital Enterprise Research Institute University of Innsbruck [email protected]
![Page 26: Ontology-Based Context Modeling · Challenges and Problems • Open topic in the Semantic Web community – combination of FOL and LP • causes undecidability, ongoing research –](https://reader033.vdocuments.us/reader033/viewer/2022050602/5fa9cb1a2e589c02fc28b4fc/html5/thumbnails/26.jpg)
making semantics real.26 3rd Workshop on Context Awareness for Proactive Systems19 June 2007
9th Int’l Conference on Ubiquitous Computing (UbiComp 2007)
16.-19. Sept 2007 in Innsbruck, Austria
www.ubicomp2007.org