Automating Semantic Metadata Collection in the Field with Mobile Application
Download as PPT, PDF4 likes821 views
The document discusses the Jefferson Project, a collaboration to create an instrumented water ecosystem, focusing on environmental monitoring and analysis through automated metadata collection using mobile applications. It highlights the shift from traditional sampling methods to a more integrated approach utilizing sensor networks for real-time data collection and hypothesis testing related to lake ecology. The proposed Mobile Context Capture for Sensor Networks aims to enhance metadata collection and improve scientific research through more effective data management and contextual understanding.
Automating Semantic Metadata Collection in the Field with Mobile Application
- 1. 1 Automating Semantic Metadata Collection in the Field with Mobile Application Laura Kinkead*, Paulo Pinheiro, Deborah L. McGuinness Tetherless World Constellation Rensselaer Polytechnic Institute * Now at Athena Health
- 2. Motivation: Next Generation Monitored Ecosystems The Jefferson Project (JP) is a joint effort between Rensselaer Polytechnic Institute (RPI), IBM and the Fund for Lake George aimed at creating an instrumented water ecosystem along with an appropriate cyberinfrastructure that can serve as a global model for ecosystem monitoring, exploration, understanding, and prediction.
- 3. 3 Historical Sampling to Sensors, Models, Experiments • Sampling at 12 locations • Only water chemistry was previously measured • Key previous results: Salt levels increasing – now dominant in the lake Chlorophyll slowly increasing Hypoxia in Caldwell Basin changed little • Key resulting hypotheses: Increasing salt levels and organic nutrients may favor dominance of cyanobacteria in the phytoplankton Ca levels may limit spread of invasive zebra mussels Chlorophyll increase may be caused by nutrient loading Food web mostly driven by “bottom-up” factors (i.e. nutrients, growing season length) Moving to sensors, streaming data, and a smarter, instrumented lake with the goal of providing a foundation to form and evaluate hypotheses much more effectively enabling a new generation of strategic science dedicated to fuller understanding of the Lake's ecological health.
- 4. 4 Science to Inform Solutions To Realize a truly Smart Lake: We need an integrative approach to understanding lake stressors, identifying correlations, hypothesizing causation, experimentally testing hypotheses, and proposing actions Science-based Solutions: Leveraging deep understanding of multiple communities and their research content to propose solutions along with evidence informs Cyberinfrastructure/Data Platform/Viz Lab Semantic Data Model Current focus has been on observations &sensor networks
- 5. 5 Traditional Data Collection Notes Notes taken in the field with the use of pen and paper Notes are rarely attached to data There is no community- wide consensus on how to take and reuse field notes
- 6. 6 Mobile Context Capture for Sensor Networks (MOCCASN) COLLECT METADATA One single mobile application capable of taking field notes and connect the notes to data as semantic annotations
- 7. SOLR-CCSVSOLR-CCSV CCSV-LoaderCCSV-Loader data Static metadata CSV2CCSV (ICS) CSV2CCSV (ICS) CCSV-Annotator*CCSV-Annotator* MOCASSNMOCASSN HASNetO-LoaderHASNetO-Loader Dynamic metadata Sensor network technician scientist data user (incl. scientists) maintains reports human Interventions (deployments, sensor config, calibrations) Single instrument data (csv) ccsv data (csv) Spreadsheet of static metadata ccsv static metadata turtle SPARQL and Lucene queries CCSV BrowserCCSV Browser SPARQL and Lucene queries Faceted search annotated csv Dynamic metadata IN-SITU DATA-SITE DATA-SITE WWW uses reports needs Dynamic metadata Ontologies (HASnetO, OBOE, PROV, VSTO) * Tool to be developed metadata metadata mainly data flow mainly metadata flow
- 9. Human-Aware Sensor Network Ontology (HASNetO) 9 vstoi: Detector vstoi: Instrument vstoi: Platform hasneto: Sensing Perspective oboe: Characteristic oboe: Entity vstoi: Detachable Detector vstoi: Attached Detector * * * 1 0..1 * hasPerspective Characteristic perspectiveOf prov: Activity hasneto: DataCollection vstoi: Deployment xsd:dateTime xsd:dateTime hasData Collection 1* prov: Agent wasAssociatedWith startedAtTime endedAtTime 1 1 * * * * oboe: Measurement of-characteristic hasneto: hasMeasurement 1 1 * * Platform 3952 Instrument 3 D 38 D 94
- 10. 10 Platform 3952 Instrument 3 D 38 D 94 RFID Type Parent Deployed Location Start Time End Time 3952 Platform NA TRUE 43.1, -73.2 2014-10- 27T12:00 NA 3 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 38 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 94 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA Example Knowledge Base
- 11. 11 Platform 3952 Instrument 5 D 74 Instrument 3 D 38 D 94 RFID Type Parent Deployed Location Start Time End Time 3952 Platform NA TRUE 43.1, -73.2 2014-10- 27T12:00 NA 3 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 38 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 94 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA Example Knowledge Base New instrument deployment
- 12. 12 Platform 3952 Instrument 5 D 74 Instrument 3 D 38 D 94 38 94 3952 74 5 3 RFID Type Parent Deployed Location Start Time End Time 3952 Platform NA TRUE 43.1, -73.2 2014-10- 27T12:00 NA 3 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 38 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 94 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA Example Knowledge Base
- 13. 13 Platform 3952 Instrument 5 D 74 Instrument 3 D 38 D 94 RFID Type Parent Deployed Location Start Time End Time 3952 Platform NA TRUE 43.1, -73.2 2014-10- 27T12:00 NA 3 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 38 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 94 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 5 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T16:30 NA 74 Detector 5 TRUE 43.1, -73.2 2014-10- 27T16:30 NA Example Knowledge Base
- 14. 14 Platform 3952 Instrument 5 D 74 Instrument 3 D 38 D 94 RFID Type Parent Deployed Location Start Time End Time 3952 Platform NA TRUE 43.1, -73.2 2014-10- 27T12:00 NA 3 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 38 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 94 Detector 3 TRUE 43.1, -73.2 2014-10- 27T12:00 NA 5 Instrument 3952 TRUE 43.1, -73.2 2014-10- 27T16:30 NA 74 Detector 5 TRUE 43.1, -73.2 2014-10- 27T16:30 NA Example Knowledge Base
- 15. 15 a more complicated example – taking an instrument out of service and adding instruments
- 16. 16 RFID Type Parent Deployed Location Start Time End Time 9754 Platform NA TRUE 43.2, -73.1 2014-10- 01T11:00 NA 8 Instrument 9754 TRUE 43.2, -73.1 2014-10- 01T11:00 NA 43 Detector 8 TRUE 43.2, -73.1 2014-10- 01T11:00 NA Platform 9754 Instrument 8 D 43 Example Knowledge Base
- 17. 17 Platform 9754 Instrument 2 D 61 Instrument 6 D 09 RFID Type Parent Deployed Location Start Time End Time 9754 Platform NA TRUE 43.2, -73.1 2014-10- 01T11:00 NA 8 Instrument 9754 TRUE 43.2, -73.1 2014-10- 01T11:00 NA 43 Detector 8 TRUE 43.2, -73.1 2014-10- 01T11:00 NA Example Knowledge Base Undeploy one instrument (8) (with one detector(43)) and deploy 2 new instruments (each with a detector)
- 18. 18 Platform 9754 Instrument 2 D 61 Instrument 6 D 09 09 61 9754 6 2 RFID Type Parent Deployed Location Start Time End Time 9754 Platform NA TRUE 43.2, -73.1 2014-10- 01T11:00 NA 8 Instrument 9754 TRUE 43.2, -73.1 2014-10- 01T11:00 NA 43 Detector 8 TRUE 43.2, -73.1 2014-10- 01T11:00 NA Example Knowledge Base
- 19. 19 Platform 9754 Instrument 2 D 61 Instrument 6 D 09 09 61 6 2 Does Detector 09 belong to Instrument 2? Yes No RFID Type Parent Deployed Location Start Time End Time 9754 Platform NA TRUE 43.2, -73.1 2014-10- 01T11:00 NA 8 Instrument 9754 TRUE 43.2, -73.1 2014-10- 01T11:00 NA 43 Detector 8 TRUE 43.2, -73.1 2014-10- 01T11:00 NA Example Knowledge Base
- 20. 20 RFID Type Parent Deployed Location Start Time End Time 9754 Platform NA FALSE 43.2, -73.1 2014-10- 01T11:00 2014-10- 27T17:00 8 Instrument 9754 FALSE 43.2, -73.1 2014-10- 01T11:00 2014-10- 27T17:00 43 Detector 8 FALSE 43.2, -73.1 2014-10- 01T11:00 2014-10- 27T17:00 9754 Platform NA TRUE 43.2, -73.1 2014-10- 27T17:00 NA 2 Instrument 9754 TRUE 43.2, -73.1 2014-10- 27T17:00 NA 6 Instrument 9754 TRUE 43.2, -73.1 2014-10- 27T17:00 NA 61 Detector 2 TRUE 43.2, -73.1 2014-10- 27T17:00 NA 9 Detector 6 TRUE 43.2, -73.1 2014-10- 27T17:00 NA Platform 9754 Instrument 2 D 61 Instrument 6 D 09 Does Detector 09 belong to Instrument 2? Yes No Example Knowledge Base
- 21. 21 RFID Type Parent Deployed Location Start Time End Time 9754 Platform NA FALSE 43.2, -73.1 2014-10- 01T11:00 2014-10- 27T17:00 8 Instrument 9754 FALSE 43.2, -73.1 2014-10- 01T11:00 2014-10- 27T17:00 43 Detector 8 FALSE 43.2, -73.1 2014-10- 01T11:00 2014-10- 27T17:00 9754 Platform NA TRUE 43.2, -73.1 2014-10- 27T17:00 NA 2 Instrument 9754 TRUE 43.2, -73.1 2014-10- 27T17:00 NA 6 Instrument 9754 TRUE 43.2, -73.1 2014-10- 27T17:00 NA 61 Detector 2 TRUE 43.2, -73.1 2014-10- 27T17:00 NA 9 Detector 6 TRUE 43.2, -73.1 2014-10- 27T17:00 NA Platform 9754 Instrument 2 D 61 Instrument 6 D 09 Example Knowledge Base Automatic update from answering one simple question: lightweight use of semantics
- 22. 22 Conclusion • Automated Metadata capture can enable current and next generation sensor-based science by enabling ubiquitous capture of contextual information – helps eliminate forgetting to annotate • Mobile technology should and can enable contextual capture even without connectivity • Relatively light weight semantics can significantly • Improve deployment quality by using semantic constraints to check for inconsistencies and help identify / resolve ambiguities • Enable integration • Enable discovery Questions? Interested in collaborating? dlm@cs.rpi.edu pinhep@rpi.edu
- 23. Extras 23
- 24. 24
- 25. 25 Recognized Challenges • What do you do when there’s no cell service? • How do you make sure the instruments on the boat are excluded?
- 26. 26 Intelligent Deployment of Sensor Networks • Automates the collection of metadata ‣ faster ‣ harder to forget to do ‣ less error-prone
- 27. The Human-Aware Sensor Network Ontology 27
- 28. Science to Inform Solutions
Editor's Notes
- #2: The ways you have worked with data in the past won’t always work for example with current and next generation smart ecosystems PUT NOTES ON SLIDES classic semantic approach, focus on metadata in the future
- #4: The 30 Year Study provides a firm foundation for identifying and responding to threats and stressors facing Lake George--including salt and invasive species--and for conducting a new generation of strategic science dedicated to fuller understanding of the Lake's ecological health.
- #5: The integrative approach to understanding and predicting requires the integration of data from multiple communities. In this talk, we will introduce a semantic data model for the jefferson project. The data model is a common infrastructure in support of data curation, integration and quality.
- #8: Where mocassn fits within a larger infrastructure
- #29: The integrative approach to understanding and predicting requires the integration of data from multiple communities. In this talk, we will introduce a semantic data model for the jefferson project. The data model is a common infrastructure in support of data curation, integration and quality.