Aggregating Linked Sensor Data

Preview:

DESCRIPTION

This presentation was given at the Semantic Sensor Networks Workshop at the International Semantic Web Conference (ISWC) 2011. It shows our approach for Aggregating Linked Sensor Data.

Citation preview

Aggregating Linked Sensor Data

Christoph Stasch, Sven Schade, Alejandro Llaves, Krysztof Janowicz, Arne Bröring

Institute for GeoinformaticsWestfälische Wilhelms-Universität Münster

3rd Workshop on Semantic Sensor NetworksBonn, 2011

Christoph Stasch – staschc@uni-muenster.de

Introduction

Christoph Stasch – staschc@uni-muenster.de

2

Aggregation in Linked Sensor Data

15°C

16°C 17°C

14°C

Adding new links:Belongs the observation value to that feature?

Spatial Aggregation

15,5°CLinking the aggregatedobservation

Christoph Stasch – staschc@uni-muenster.de

3

Spatio-temporal and Thematic Aggregation

Christoph Stasch – staschc@uni-muenster.de

4

Aggregation• Aggregation:

• An aggregation process computes a value, an aggregate, for a group of attribute values by means of an aggregation function. The attribute values are grouped by a partitioning predicate.

• Aggregation Function: • Function used to compute the aggregate.

• Partitioning Predicate: • Predicate used to group objects before aggregating the

values attached to these objects.

Christoph Stasch – staschc@uni-muenster.de

5

Spatio-temporal vs. Thematic Aggregation

• Spatio-temporal Aggregation: – Partitionining predicate is spatial and/or temporal

• Thematic Aggregation:– Partitioning predicate operates on attribute values

Christoph Stasch – staschc@uni-muenster.de

6

Previous Work

Christoph Stasch – staschc@uni-muenster.de

7

Linked Sensor Data• World Wide Web is for websites /

documents– HTTP– HTML– ...

• Sensor Web is for sensors– SOS– O&M– ...

• Linked Data Web is for linked data– RDF

• Linked Sensor Data (e.g. Page 2009)

Christoph Stasch – staschc@uni-muenster.de

8

RESTful SOS Proxy• Proxy service for Sensor Observation Services• Linked data model + URI scheme for observation

resources

Christoph Stasch – staschc@uni-muenster.de

9

Janowicz, K., Bröring, A., Stasch, C., Schade, S., Everding, T., and Llaves, A. (2011): A RESTful Proxy and Data Model for Linked Sensor Data. International Journal of Digital Earth. DOI:10.1080/17538947.2011.614698, pp. 1-22

Spatio-Temporal Aggregation Service(STAS)

Christoph Stasch – staschc@uni-muenster.de

10

Stasch, C., Autermann, C., Foerster, T., Pebesma, E.: Towards a Spatiotemporal Aggregation Service in the Sensor Web. Poster Presentation. In: The 14th AGILE International Conference on Geographic Information Science. (2011)

Aggregating Linked Sensor Data

Christoph Stasch – staschc@uni-muenster.de

11

Aggregating Linked Sensor Data

• Linked Data Model:– Extending the SSO pattern to allow aggregated

observations

• Effects on Links from and To Observations– How do links change during aggregation?

• Provenance– Information is contained in Linked Data Model; can be

mapped to Open Provenance Model or Provenance Vocabulary

Christoph Stasch – staschc@uni-muenster.de

12

Extended SSO Design Pattern

Christoph Stasch – staschc@uni-muenster.de

13

Effects on Links from and to Observations

Christoph Stasch – staschc@uni-muenster.de

14

15°C

16°C 17°C

14°CFOI1

FOI2

FOI3

FOI4

Spatial Aggregation

15,5°C

Effects from and to Observations

Christoph Stasch – staschc@uni-muenster.de

15

Provenance

Christoph Stasch – staschc@uni-muenster.de

16

Provenance Information• Common approaches:

– Open Provenance Model• Nodes and edges to define provenance graphs

– Provenance Vocabulary

• Provenance in Sensor Data:– Information about the source of the data as well as

transformations applied– Approaches

• Provenance in Linked Sensor Data• Using OPM for sensor data• Defining own provenance models

Christoph Stasch – staschc@uni-muenster.de

17

Provenance

Christoph Stasch – staschc@uni-muenster.de

18

DUL = Dolce Ultra Lightldm = Linked Sensor Data Modelopmv = Open Provenance Model Vocabularyprv = Provenance Vocabulary

Conclusions & Outlook

Christoph Stasch – staschc@uni-muenster.de

19

Conclusions• Aggregation helps:

– Establishing new links– Fusing datasets

• Extended SSO pattern– Allows for aggregated observations and aggregation

processes– Retracing aggregated Observations back to original

observations mapping to OPM and Provenance Vocabulary

• Effects of aggregation on links from and to observations

Christoph Stasch – staschc@uni-muenster.de

20

Outlook• Formalize effects of aggregation on links• Enable Spatio-temporal Aggregation Service for

linked sensor data• Integrate with approaches for sensor plug‘n‘play

and linked sensor streams• Utilize semantics of aggregation processes• Integrate uncertainty/quality information

Christoph Stasch – staschc@uni-muenster.de

21

Discussion

Christoph Stasch – staschc@uni-muenster.de

22

Discussion• To what aggregation level can we speak of

observations?• Virtual sensors vs. Physical Sensors?• Common aggregation mechanisms in Linked

Data?

Christoph Stasch – staschc@uni-muenster.de

23

Thank you!

RESTful SOS:http://52north.org/communities/sensorweb/clients/OX_RESTful_SOS/index.htm

STAS:

https://wiki.aston.ac.uk/foswiki/bin/view/UncertWeb/Spatio-temporalAggregationService

http://www.uncertweb.org

http://www.envirofi.org

http://www.envision-project.eu

http://irtg.ifgi.de

Christoph Stasch – staschc@uni-muenster.de

24

Recommended