Harshal Patni, "Real Time Semantic Analysis of Streaming Sensor Data," MS Thesis Defense, Kno.e.sis Center, Wright State University, Dayton OH, March 21, 2001.
More at: http://wiki.knoesis.org/index.php/SSW
Dissertation Advisor: Prof. Amit Sheth
Real Time Semantic Analysis of Streaming Sensor Data
1.
2. x WEB DATA evolved over time Real-Time Sensor, Social, Multi-media data 2010’s Dynamic User Generated Content 2000’s Static Document and files 1990’s 2
3. x Properties of Streaming Data Huge Volume Rapid Continuous Information Overload!! Heterogeneous 3
4. x Some Statistics “A cross-country flight from New York to Los Angeles on a Boeing 737 plane generates a massive 240 terabytes of data” - GigaOmni Media “Sensors Networks will produce 10-20 times the amount of generated by social media in the next few years” - GigaOmni Media “More data has been created in the last three years than in all the past 40,000 years” - Teradata Solution - “Meaningfully summarize this data” 4
5. 48th ACM Southeast Conference. ACMSE 2010. Oxford, Mississippi. April 15-17, 2010. From Sensor Streams to Feature Streams in Real Time HarshalPatni Ohio Center of Excellence in Knowledge enabled Computing (Kno.e.sis) Wright State University, Dayton, OH Part of Semantic Sensor Web @ Kno.e.sis
6. x Outline Introduction Architecture Linked Sensor Data Feature Streams Demonstration 6
7. x Domain Weather Domain Features Blizzard Flurry RainStorm RainShower 7
8. x Explaining the title Background Knowledge Blizzard Rain Storm ABSTRACTION Huge amount of Raw Sensor Data Features representing Real-World events 8
9. x Types of Abstractions Summarization across Thematic Dimension Summarization over the Temporal Dimension 9
10. x Types of Abstractions Summarization across Thematic Dimension Select Join Background Knowledge Analyze Features representing Real-World Events 10
11. x An example problem? 11 “Find the sequence of weather events observed near Dayton James Cox Airport between Jan 13th and Jan 18th?” Spatial Thematic Temporal Technologies required - Linked Sensor Data Feature Streams
15. 48th ACM Southeast Conference. ACMSE 2010. Oxford, Mississippi. April 15-17, 2010. Technology1: Linked Sensor Data Find the sensor around Dayton James Cox Airport? Extract Data for the sensor near Dayton James Cox Airport? Harshal Patni, Cory Henson, Amit Sheth, 'Linked Sensor Data,' In: Proceedings of 2010 International Symposium on Collaborative Technologies and Systems (CTS 2010), Chicago, IL, May 17-21, 2010.
16. Sensor Discovery Application Weather Station ID Current Observations from MesoWest Weather Station Coordinates Weather Station Phenomena MesoWest – Project under Department of Meteorology, University of UTAH GeoNames – Geographic dataset 16
17. What is Linked Sensor Data Weather Sensors Sensor Dataset GPS Sensors Satellite Sensors Camera Sensors 17
18. What is Linked Sensor Data Recommended best practice for exposing, sharing, and connecting pieces of data, information, and knowledge on the Web using URIs and RDF GeoNames Dataset RDF – language for representing data on the Web locatedNear Sensor Dataset Publicly Accessible 18
19. Linked Sensor Data on LOD - First Sensor Dataset on LOD - Among the largest dataset on LOD 19
40. 48th ACM Southeast Conference. ACMSE 2010. Oxford, Mississippi. April 15-17, 2010. Technology 2: Feature Streams What feature is currently being detected by sensor near Dayton Airport? Harshal Patni, Cory Henson, Amit Sheth, Pramod Ananthram, ‘From Real Time Sensor Streams to Real Time Feature Streams,' Kno.e.sis Technical Report, January 2011.
41. x System Architecture Streams Integration based on feature composition Integrated Stream Analysis to check if the feature is being detected 30
45. x System Architecture Integrated Stream Analysis to check if the feature is being detected 34
46. x Feature Definition RainStorm = HighWindSpeed(above 35mph) AND Rain Precipitation AND Temperature(greater than 32F) SPARQL query for RainStorm Temperature Rain Precipitation WindSpeed 35 Rain Storm NOAA definition
58. x Outline Introduction Architecture Linked Sensor Data Feature Streams Demonstration 40
59. x Demo 41 Feature Streams Demo http://knoesis1.wright.edu/EventStreams
60.
61. WORKSHOP PAPERS Harshal Patni, Satya S. Sahoo, Cory Henson, Amit Sheth, Provenance Aware Linked Sensor Data, 2nd Workshop on Trust and Privacy on Social and Semantic Web,Co-Located with ESWC, Heraklion Greece, May 30th - June 3rd 2010 Harshal Patni, Cory Henson, Amit Sheth, Linked Sensor Data, In: Proceedings of 2010 International Symposium on Collaborative Technologies and Systems (CTS 2010), Chicago, IL, May 17-21, 2010 TECHNICAL REPORT Harshal Patni, Cory Henson, Amit Sheth, and Pramod Ananthram. From Real Time Sensor Streams to Real Time Feature Streams, Kno.e.sis Center Technical Report, December 2009 Joshua Pschorr, Cory Henson, Harshal Patni, and Amit Sheth. Sensor Discovery on Linked Data, Kno.e.sis Center Technical Report, December 2009 JOURNAL PAPER (In Progress) Semantic Sensor Web: Design and Application towards weaving a meaningful sensor web Publications 43
64. Demos, Papers and more at: http://wiki.knoesis.org/index.php/SSW Semantic Sensor Web @ Kno.e.sis QUESTIONS 46
Notes de l'éditeur
Good Morning Everyone. My name is Harshal Patni and I am here to present my thesis on Streaming Sensor Data but Before we begin lets have a look at how web data evolved over time
Social media is the dominant source of streaming data now, however in future sensors would …Data needs to be reduced
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
Move this slide above
Remove the precipitation (in) and also show the general streamAdd the image taken on the phoneRemove the stuff on left when you show select, join and analyze`
Remove the precipitation (in) and also show the general streamAdd the image taken on the phoneRemove the stuff on left when you show select, join and analyze`
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsAdd linked Sensor Data when highlightThe output of these phases is called LSD and its added on LOD
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
Get all sensors using well known location names – Problem to be solveAssociate sensor descriptions to well know locations.
Get all sensors using well known location names – Problem to be solve
Say the numbers in the table
RDF because of LOD
Highlight the important points in MesoWest DataThe sensor data file just 3 linesMapping file - shorten
Emphasize semantically annotated O&MAnd its an XMLTry to replace the cory/weather.owl
Use the ssn ontologyAdd the image of ontology for the (Sensor Ontology)http://www.w3.org/2005/Incubator/ssn/wiki/Report_Work_on_the_SSN_ontology
Add in block letters saying this is semantically annotated XML and RDF
Add Pubby to show derefenced dataPubby should be large to show what it is
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
Replace Air Temperature with Non Freezing Temperature
Replace Rain Precipitation with PrecipitationSame with airtempearure - temperature
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
Highlight the query with 3 boxes to show the temp,windspeed and precipitation streamHighlight the feature results too
Talk about the observations and features storage
Remove the precipitation (in) and also show the general streamAdd the image taken on the phoneRemove the stuff on left when you show select, join and analyze`
Linked Data explodes
To walk through the implementation, lets take a sample questionThe question might look trivial, but it contains 3 important partsMarket the datasets we added on LOD
Linked Data explodes
% of FeaturesThrow the text on the top for the statisticsMiddle of storm and hence we have 70 % data reductionElse it would be more