WCS Technical Description

From OPeNDAP Documentation
Jump to: navigation, search

1 How the WCS service works

Components of the WCS Service

Just enough information to be dangerous ...

1.1 WCS

The WCS 1.1.2 software implements a WCS service that is close to generic. It can work using a simple XML catalog that lists WCS coverages ('coverages' hereafter), at lest in theory. This software using the semantic.wcs package to perform actions like access coverages and rebuild the catalog of served coverages.

1.2 semantics.wcs

The software in semantic.wcs is the interface to the 'semantic engine' that the IRISail software implements. It doesn't know anything about inferencing per se, but it implements an interface to the high-level operations provided by the inferencing software. All of the code is contained in one class called StaticRdfCatalog that has a number of public methods that return coverages and other useful information about coverages. It also contains an update() method used to (re)build the contents of the semantic repository and then the coverages that can be derived from the information in that repository. The end result of running the update() method is that an XML file listing the known coverage descriptions is written into the content/WCS/StaticRdfCatalog/coverageXMLfromRDF.xml file. This file is then used to build an in-memory representation of WCS coverages that the other methods of the StaticRdfCatalog class use.

1.3 IRISail

This is the semantic engine. It uses information about WCS and other metadata standards to build coverages for WCS using whatever metadata is available. It leverages ontologies that describe the different metadata standards & conventions in general terms.

The entry point for the inferencing software is the class RespoitoryOps.

The inferencing software uses the sesame software for both inferencing and as a tripple store for information, both initial conditions and derived information.

1.4 Important classes

These are all .java classes and can be found in the olfs/src/opendap/{semantics,wcs} directories.

1.4.1 StaticRdfCatalog

The interface used to access coverages and to build coverages using the semantic engine.

This class is used to retrieve coverage descriptions of a data set. A Sesame-OWLIM RDF store is initialized and populated and an xml document of coverages of the served data sets is generated. This xml document is used to answer request posted by a client. The RDF store is updated periodically to make sure the information served is up to date. Care is taken to make sure the RDF store can be used by multiple users at the same time.
The RDF store is populated by importing and ingesting semantic inference rules and owl/xml schema files and rdf files describing data sets. The doNotImportTheseUrls is a string vector that holds URL of bad files that should not be imported. The catalogCacheDirectory holds the OWLIM persistent directory owlim_storage_folder, the coverage description document and the dump of the whole repository files in triple graph format.

1.4.2 RepositoryOps

The interface to the semantic engine.

This class is the major class that manipulates and maintains the repository up to date. Using this class StartingPoint statements are introduced in the repository; documents or data sets no longer to be served anymore is deleted from the repository.
The repository consists of StartingPoint RDF document and RDF documents needed by those StartingPoints. When introducing a StartingPoint into the repository, two things need to be added. One is the StartingPoint statement and the other is the content of the StartingPoint RDF document. e.g. "x rdf:type rdfcache:StartingPoint" is a StartingPoint statement. "x" is the URL of the RDF document of a StartingPoint. The startingpoint statement is added using this class then the actual content is added later using class RdfImporter.
rdfcache = <http://iridl.ldeo.columbia.edu/ontologies/rdfcache.owl#>

1.4.3 CoverageDescription

The interface to the collection of coverages.

An implementation of a wcs:CoverageDescription object. This implementation includes methods that assist in the creation of DAP constraint expressions to retrieve coverage data as NetCDF.

1.4.4 ConstructRuleEvaluator

This class is used to run external inference rules on an Sesame-OWLIM repository. The inference rules are written in SeRQL. The inferred statements are added into the repository. The inference rules are run repeatedly until no rules generate any new statements.