Ubiquitous Service Capability Modeling and Similarity Based Searching
1 like612 views
This document discusses modeling sensor capabilities and similarity-based searching. It proposes a capability meta-model to describe sensor capabilities using action verbs and attributes. Extended relations are defined between capabilities to capture similarity. A capability graph is built using these relations. An algorithm is presented to find similar sensor capabilities based on calculating capability distance and performing heuristic searching using simulated annealing. Future work includes completing the algorithms and evaluating performance.
![Concrete Sensor Capability
Digital Enterprise Research Institute www.deri.ie
sc:Sensor1 a ssn:SensingDevice,sc:SensorCapability;
sc:hasMeasurementFunction sc:Thermal;
sc:hasMeasurementProperty sc:Mp_1;
sc:hasObservation sc:Ob_1.
sc:Ob_1 a ssn:Observation,cap:FetchedDynamicValue;
sc:hasEndpoint "www.deri.org/sensor1"^^xsd:URI;
sc:hasLifting "www.deri.org/lifting.xsd"^^xsd:URI];
sc:hasValue “?temp”.
sc:Mp_1 a sc:Accuracy, cap:ConditionalValue;
cap:hasCondition [ cap:hasElement "0.8"^^xsd:decimal;
cap:hasExpression [ cap:hasType “SPARQL”
cap:exprValue “Filter(?temp<10&&?temp>0)”]]，
[ cap:hasElement "0.9"^^xsd:decimal;
cap:hasExpression [ cap:hasType “SPARQL”
cap:exprValue “Filter(?temp>10)”]].
Enabling Networked Knowledge
11](https://image.slidesharecdn.com/wce2012-121203124311-phpapp02/85/Ubiquitous-Service-Capability-Modeling-and-Similarity-Based-Searching-11-320.jpg)
![Concrete Sensor Capability
Digital Enterprise Research Institute www.deri.ie
sc:Sensor1 a ssn:SensingDevice,sc:SensorCapability;
sc:hasMeasurementFunction sc:Thermal;
sc:hasMeasurementProperty sc:Mp_1;
sc:hasObservation sc:Ob_1.
sc:Ob_1 a ssn:Observation,cap:FetchedDynamicValue;
sc:hasEndpoint "www.deri.org/sensor1"^^xsd:URI;
sc:hasLifting "www.deri.org/lifting.xsd"^^xsd:URI];
sc:hasValue “?temp”.
sc:Mp_1 a sc:Accuracy, cap:ConditionalValue;
cap:hasCondition [ cap:hasElement "0.8"^^xsd:decimal;
cap:hasExpression [ cap:hasType “SPARQL”
cap:exprValue “Filter(?temp<10&&?temp>0)”]]，
[ cap:hasElement "0.9"^^xsd:decimal;
cap:hasExpression [ cap:hasType “SPARQL”
cap:exprValue “Filter(?temp>10)”]].
Enabling Networked Knowledge
12](https://image.slidesharecdn.com/wce2012-121203124311-phpapp02/85/Ubiquitous-Service-Capability-Modeling-and-Similarity-Based-Searching-12-320.jpg)
![Concrete Sensor Capability
Digital Enterprise Research Institute www.deri.ie
sc:Sensor1 a ssn:SensingDevice,sc:SensorCapability;
sc:hasMeasurementFunction sc:Thermal;
sc:hasMeasurementProperty sc:Mp_1;
sc:hasObservation sc:Ob_1.
sc:Ob_1 a ssn:Observation,cap:FetchedDynamicValue;
sc:hasEndpoint "www.deri.org/sensor1"^^xsd:URI;
sc:hasLifting "www.deri.org/lifting.xsd"^^xsd:URI];
sc:hasValue “?temp”.
sc:Mp_1 a sc:Accuracy, cap:ConditionalValue;
cap:hasCondition [ cap:hasElement "0.8"^^xsd:decimal;
cap:hasExpression [ cap:hasType “SPARQL”
cap:exprValue “Filter(?temp<10&&?temp>0)”]]，
[ cap:hasElement "0.9"^^xsd:decimal;
cap:hasExpression [ cap:hasType “SPARQL”
cap:exprValue “Filter(?temp>10)”]].
Enabling Networked Knowledge
13](https://image.slidesharecdn.com/wce2012-121203124311-phpapp02/85/Ubiquitous-Service-Capability-Modeling-and-Similarity-Based-Searching-13-320.jpg)
Ubiquitous Service Capability Modeling and Similarity Based Searching
- 1. Digital Enterprise Research Institute www.deri.ie Ubiquitous Service Capability Modeling and Similarity Based Searching Feng Gao and Wassim Derguech Engineering the Semantic Enterprise (ESE2012) Workshop in WISE 2012 – Paphos, Cyprus © Copyright 2011 Digital Enterprise Research Institute. All rights reserved. Enabling Networked Knowledge
- 2. Agenda Digital Enterprise Research Institute www.deri.ie Motivation and objectives Capability meta-model Sensor capability model Extended capability relations and distance Searching algorithm Conclusions Enabling Networked Knowledge 2
- 3. Motivation & Objectives Digital Enterprise Research Institute www.deri.ie Sensor nodes have very dynamic features. describe such feature offline and discover them online. The number of similar sensor nodes in a sensor network can be huge. capture relations between sensor capabilities and to use these relations to find specified replacement capabilities among a large number of similar ones in time. (for both design time sensor re-discovery or run time system recovery) Enabling Networked Knowledge 3
- 4. Capability Meta-Model Digital Enterprise Research Institute www.deri.ie Capability = ActionVerb + {(att1,val1), (att2,val2)…} Action verb capability attribute ontology ontology Enabling Networked Knowledge 4
- 5. Capability Meta-Model Digital Enterprise Research Institute www.deri.ie Capability = ActionVerb + {(att1,val1), (att2,val2)…} Action verb capability attribute ontology ontology Capability relations: isVariantOf: subClassOf or instanceOf. Extend: has more attributes. Specify: has same attributes but at least one isVariantOf relation for attribute values. Enabling Networked Knowledge 5
- 6. Capability Meta-Model Digital Enterprise Research Institute www.deri.ie Capability = ActionVerb + {(att1,val1), (att2,val2)…} Action verb capability attribute ontology ontology Capability relations: isVariantOf: subClassOf or instanceOf. Extend: has more attributes. Specify: has same attributes but at least one isVariantOf relation for attribute values. Use these relations to build a hierarchy of capabilities. Enabling Networked Knowledge 6
- 7. Top Sensor Capability Digital Enterprise Research Institute www.deri.ie cap:Capability rdfs:subclassof sc:Sensor Capability Enabling Networked Knowledge 7
- 8. Top Sensor Capability Digital Enterprise Research Institute www.deri.ie cap:Capability rdfs:subclassof sc:Sensor sc:hasSurvivalPropterty Capability sc:hasMeasurementFunction Enabling Networked Knowledge 8
- 9. Top Sensor Capability Digital Enterprise Research Institute www.deri.ie cap:Capability ssn:Observation rdfs:subclassof sc:Sensor sc:hasSurvivalPropterty ssn:Survival Capability Property sc:hasMeasurementFunction ssn:Operational Property sc:Measurement Function ssn:Measurement Property Enabling Networked Knowledge 9
- 10. Top Sensor Capability Digital Enterprise Research Institute www.deri.ie cap:Capability ssn:Observation rdfs:subclassof sc:Sensor sc:hasSurvivalPropterty ssn:Survival Capability Property sc:hasMeasurementFunction ssn:Operational Property sc:Measurement Function rdfs:subclassof rdfs:subclassof ssn:Measurement Property rdfs:subclassof rdfs:subclassof ssn:Property Enabling Networked Knowledge 10
- 11. Concrete Sensor Capability Digital Enterprise Research Institute www.deri.ie sc:Sensor1 a ssn:SensingDevice,sc:SensorCapability; sc:hasMeasurementFunction sc:Thermal; sc:hasMeasurementProperty sc:Mp_1; sc:hasObservation sc:Ob_1. sc:Ob_1 a ssn:Observation,cap:FetchedDynamicValue; sc:hasEndpoint "www.deri.org/sensor1"^^xsd:URI; sc:hasLifting "www.deri.org/lifting.xsd"^^xsd:URI]; sc:hasValue “?temp”. sc:Mp_1 a sc:Accuracy, cap:ConditionalValue; cap:hasCondition [ cap:hasElement "0.8"^^xsd:decimal; cap:hasExpression [ cap:hasType “SPARQL” cap:exprValue “Filter(?temp<10&&?temp>0)”]]， [ cap:hasElement "0.9"^^xsd:decimal; cap:hasExpression [ cap:hasType “SPARQL” cap:exprValue “Filter(?temp>10)”]]. Enabling Networked Knowledge 11
- 12. Concrete Sensor Capability Digital Enterprise Research Institute www.deri.ie sc:Sensor1 a ssn:SensingDevice,sc:SensorCapability; sc:hasMeasurementFunction sc:Thermal; sc:hasMeasurementProperty sc:Mp_1; sc:hasObservation sc:Ob_1. sc:Ob_1 a ssn:Observation,cap:FetchedDynamicValue; sc:hasEndpoint "www.deri.org/sensor1"^^xsd:URI; sc:hasLifting "www.deri.org/lifting.xsd"^^xsd:URI]; sc:hasValue “?temp”. sc:Mp_1 a sc:Accuracy, cap:ConditionalValue; cap:hasCondition [ cap:hasElement "0.8"^^xsd:decimal; cap:hasExpression [ cap:hasType “SPARQL” cap:exprValue “Filter(?temp<10&&?temp>0)”]]， [ cap:hasElement "0.9"^^xsd:decimal; cap:hasExpression [ cap:hasType “SPARQL” cap:exprValue “Filter(?temp>10)”]]. Enabling Networked Knowledge 12
- 13. Concrete Sensor Capability Digital Enterprise Research Institute www.deri.ie sc:Sensor1 a ssn:SensingDevice,sc:SensorCapability; sc:hasMeasurementFunction sc:Thermal; sc:hasMeasurementProperty sc:Mp_1; sc:hasObservation sc:Ob_1. sc:Ob_1 a ssn:Observation,cap:FetchedDynamicValue; sc:hasEndpoint "www.deri.org/sensor1"^^xsd:URI; sc:hasLifting "www.deri.org/lifting.xsd"^^xsd:URI]; sc:hasValue “?temp”. sc:Mp_1 a sc:Accuracy, cap:ConditionalValue; cap:hasCondition [ cap:hasElement "0.8"^^xsd:decimal; cap:hasExpression [ cap:hasType “SPARQL” cap:exprValue “Filter(?temp<10&&?temp>0)”]]， [ cap:hasElement "0.9"^^xsd:decimal; cap:hasExpression [ cap:hasType “SPARQL” cap:exprValue “Filter(?temp>10)”]]. Enabling Networked Knowledge 13
- 14. Extended Capability Relations Digital Enterprise Research Institute www.deri.ie specify, extend ⊂(C × C) extended relations ⊂(C× C × A) For capabilities C1, C2 and attribute a in (A1 U A2) a ϵ (A1∩A2) C1(a)=C2(a) ShareSame C1(a)≠C2(a) ShareDiffer C1(a) isVariantOf C2(a) Specify C2(a) isVariantOf C1(a) Generalize Enabling Networked Knowledge 14
- 15. Extended Capability Relations Digital Enterprise Research Institute www.deri.ie specify, extend ⊂(C × C) extended relations ⊂(C× C × A) For capabilities C1, C2 and attribute a in (A1 U A2) a ϵ (A1∩A2) a ϵ (A1/A2) a ϵ (A2/A1) C1(a)=C2(a) ShareSame C1(a)≠C2(a) ShareDiffer C1(a) isVariantOf C2(a) Specify C2(a) isVariantOf C1(a) Generalize DifferMore DifferLess Enabling Networked Knowledge 15
- 16. Capability Graph Digital Enterprise Research Institute www.deri.ie Extended relations can be used to create a capability graph sc:TopSensor Capability Enabling Networked Knowledge 16
- 17. Capability Graph Digital Enterprise Research Institute www.deri.ie Extended relations can be used to create a capability graph sc:TopSensor Capability Secify a, b, c sc:Sensor0 Enabling Networked Knowledge 17
- 18. Capability Graph Digital Enterprise Research Institute www.deri.ie Extended relations can be used to create a capability graph sc:TopSensor Capability Secify a, b, c sc:Sensor0 Extend d Specify a; Extend f sc:Sensor2 Extend e sc:Sensor3 sc:Sensor1 Enabling Networked Knowledge 18
- 19. Capability Graph Digital Enterprise Research Institute www.deri.ie Extended relations can be used to create a capability graph sc:TopSensor Capability Secify a, b, c sc:Sensor0 Extend d Specify a; Extend f sc:Sensor2 Extend e sc:Sensor3 Extend e; sc:Sensor1 DifferLess d; Enabling Networked Knowledge 19
- 20. Capability Distance Digital Enterprise Research Institute www.deri.ie Distance in attributes: DA= (DL+DM)/(SS+SD+SP+GE) Distance in values: DV=(SD+0.5*SS+0.5*GE)/SS Enabling Networked Knowledge 20
- 21. Capability Distance Digital Enterprise Research Institute www.deri.ie Distance in attributes: DA= (DL+DM)/(SS+SD+SP+GE) Distance in values: DV=(SD+0.5*SS+0.5*GE)/SS Capability Distance: D2=DA2+DV2 D DA 0 DV Enabling Networked Knowledge 21
- 22. Heuristic Searching Digital Enterprise Research Institute www.deri.ie Goal: find suitable replacement for a certain sensor capability among a large sensor network in time. Approach: Simulated Annealing based on capability distance. Guarantee to finish in given time span. Cannot guarantee a complete result set. Enabling Networked Knowledge 22
- 23. How does it work Digital Enterprise Research Institute www.deri.ie sc:target a sc:cap; sc:hasMp … sc:hasOp … … Enabling Networked Knowledge 23
- 24. How does it work Digital Enterprise Research Institute www.deri.ie sc:target a sc:cap; t=5 t=3 1. Set initial distance d as temperature, sc:hasMp … add current node into accepted nodes. sc:hasOp … … Enabling Networked Knowledge 24
- 25. How does it work Digital Enterprise Research Institute www.deri.ie sc:target a sc:cap; t=5 t=3 1. Set initial distance d as temperature, sc:hasMp … add current node into accepted nodes. sc:hasOp … 2. Get distance d’ of unvisited neighbors … of current node. d=6 d = 10 d=3 Enabling Networked Knowledge 25
- 26. How does it work Digital Enterprise Research Institute www.deri.ie sc:target a sc:cap; t=5 t=3 1. Set initial distance d as temperature, sc:hasMp … add current node into accepted nodes. sc:hasOp … 2. Get distance d’ of unvisited neighbors … of current node. d=6 3. If d’<t, accept neighbor, else accept with a probability P(t), label neighbor as d = 10 visited, remove current node from d=3 accepted nodes. 4. Repeat 2-3 for all accepted nodes for several iterations. Enabling Networked Knowledge 26
- 27. How does it work Digital Enterprise Research Institute www.deri.ie sc:target a sc:cap; t=5 t=3 1. Set initial distance d as temperature, sc:hasMp … add current node into accepted nodes. sc:hasOp … 2. Get distance d’ of unvisited neighbors … of current node. d=6 3. If d’<t, accept neighbor, else accept with a probability P(t), label neighbor as d = 10 visited, remove current node from d=3 accepted nodes. 4. Repeat 2-3 for all accepted nodes for several iterations. 5. Decrease temperature d according to d=2 schedule, repeat 2-5 till temperature falls to zero or no more unvisited nodes. 6. Return accepted nodes as results. Enabling Networked Knowledge 27
- 28. Conclusions & Future Work Digital Enterprise Research Institute www.deri.ie In this paper we: demonstrated how sensor capabilities can be modeled in a dynamic and related way, proposed an heuristic search algorithm to find replacement for sensor capabilities. Future work: Complete algorithms and test the performance. Develop algorithms for soft constraints or partial matches. Enabling Networked Knowledge 28