1 / 1

Semantic Representation of Temporal Metadata in a Virtual Observatory

IN31B-1444. Temporal Metadata Modeling. Semantic Representation of Temporal Metadata in a Virtual Observatory Han Wang 1 ( wangh17@rpi.edu ) , Eric Rozell 1 ( rozele@ rpi.edu ) , Patrick West 1 ( westp@rpi.edu ) , Stephan Zednik 1 ( zednis@rpi.edu ) ,

brinly
Download Presentation

Semantic Representation of Temporal Metadata in a Virtual Observatory

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. IN31B-1444 Temporal Metadata Modeling Semantic Representation of Temporal Metadata in a Virtual Observatory Han Wang1 (wangh17@rpi.edu), Eric Rozell1 (rozele@rpi.edu), Patrick West1 (westp@rpi.edu), Stephan Zednik1 (zednis@rpi.edu), Peter Arthur Fox1 (pfox@cs.rpi.edu) 1Rensselaer Polytechnic Institute, 110 8th St., Troy, NY, United States Abstract The Virtual Solar-Terrestrial Observatory (VSTO) Portal at vsto.org provides a set of guided workflows to implement use cases designed for solar-terrestrial physics and upper atmospheric science. Semantics are used in VSTO to model abstract instrument and parameter classifications, providing data access to users without extended domain specific vocabularies. The temporal restrictions used in the workflows are currently possible via RESTful services made to a remote system with access to a SQL-based metadata catalog. In order to provide a greater range of temporal reasoning and search capabilities for the user, we propose an alternative architecture design for the VSTO Portal, where the temporal metadata is integrated in the domain ontology. We achieve this integration by converting temporal metadata from the headers of raw data files into RDF using the OWL-Time vocabulary. This presentation covers our work with semantic temporal metadata, including: our representation using OWL-Time, issues that we have faced in persistent storage, and performance and scalability of semantic query. We conclude with discussions of the significance semantic temporal metadata has in virtual observatories. Fig. 1. XML Schema DatatypedateTime Strings Fig. 2. Verbose OWL-Time Instances Fig. 3. Discrete Interval Coverage in OWL-Time Fig. 1 shows a model that represents the start and end times for VSTO dataset records using only xsd:dateTime strings. Fig. 2 shows a model that represents the start and end times for VSTO dataset records using OWL-Time instances with a granularity of seconds. Fig. 3 show a model that represents the individual observations of VSTO dataset records using a notion of discrete intervals created within the SeSF ontology. All the three models above are not feasible solutions as they require the SPARQL engine to parse at least O(106) time instances to answer the queries for VSTO workflow with interactive responses (query response in less than 10 seconds). Fig. 4 illustrates a feasible solution for the temporal metadata modeling. It represents the start and end times for a VSTO dataset record using xsd:dateTime strings with a granularity of seconds, and it also includes the exact temporal range for that dataset to a granularity of days (as required by the use cases) using time:DateTimeInterval. This modeling solution only increases the size of the data by approximately a factor of 5. We have achieved interactive response time for all scenarios in the original VSTO workflow by answering various SPARQL queries, which are described in more detail below. Fig. 4. OWL-Time Instances with Date Coverages Motivations and Use Cases Temporal Model Evaluation • Limitations of relational database representations: • There are no mechanisms for inferring relationships given relationships that exit within the relational database. While with an ontology, we can add inferencing to the temporal instances. • There are no easy ways of inheriting relationships, as can be done with an ontology and knowledge base. • The expressivity of a temporal model represented by a relational database schema is much smaller than an ontological temporal model. • Use case #1: Retrieve data where two or more instruments have coincident measurements within a temporal interval. • Example: Retrieve any data where the Millstone Hill Fabry Perot Interferometer and the Poker Flat Fabry Perot Interferometer and collecting data simultaneously. • Use case #2: Retrieve data in a non-contiguous time interval. • Example: Retrieve data for sunspot activity during the month of March (in any year). Table 1 lists 12 workflow tasks derived from the VSTO Portal. We generated SPARQL queries representing these tasks and executed them in a Virtuoso tripe store loaded with CEDAR datasets, which have approximately 20 million time instances (about 80 million triples). The average execution time of these queries all fall within 1 second, which either improves or is comparable to the task performance using RESTful service calls. The following shows a SPARQL query for Task 1, which takes about 5 seconds with the RESTful service. Instrument 175 is a instrument class called Space Craft, and it has the highest number of days covered. Table 1. A table of workflow tasks derived from the VSTO Portal PREFIX vsto: <http://escience.rpi.edu/ontology/vsto/2/0/vsto.owl#> PREFIX cedar: <http://escience.rpi.edu/ontology/vsto/2/0/cedar.owl#> PREFIX time: <http://www.w3.org/2006/time#> PREFIX xsd: <http://www.w3.org/2001/XMLSchema#> SELECT DISTINCT ?year WHERE { ?dataset vsto:isFromInstrument cedar:cedar_instrument_175 . ?dataset vsto:hasDateTimeCoverage ?interval . ?interval time:hasDateTimeDescription ?desc . ?desctime:year ?year . } Research Methodology Visit our working group page at http://goo.gl/q9mmn Get the poster at http://goo.gl/M2UkF • Design and implementation of temporal models for ontologies • Use OWL-Time as a start point • Keep the total number of triples over time instances small • Evaluation of performance and scalability of Semantic Web tools, in particular, scalable storage and SPARQL querying. • Load triples into Virtuoso triple store • Generate SPARQL queries based on recurring tasks in the workflow Glossary: RPI – Rensselaer Polytechnic Institute TWC – Tetherless World Constellation at Rensselaer Polytechnic Institute VSTO –Virtual Solar-Terrestrial Observatory CEDAR – Coupling Energetics and Dynamics of Atmospheric Regions SeSF – Semantic eScience Framework RESTful– Representational State Transfer Sponsors: NSF Office of Cyberinfrastructure (OCI) References: P. West, E. Rozell, S. Zednik, P. Fox, and D. McGuinness, 2009, Semantically Enabled Temporal Reasoning in a Virtual Observatory, OWL Experiences and Directions, CEUR Workshop Proceedings, vol. 529. E. Rozell, P. West, and P. Fox, 2010, Experiences Integrating Temporal Metadata in a Domain Ontology, Technical Report.

More Related