Kepler Meeting SMS Notes

This is version 9. It is not the current version, and thus it cannot be edited.
[Back to current version]   [Restore this version]

The Semantic Mediation System and KEPLER

Back to Kepler Meeting Agenda

"Eco" ontologies

• In SEEK we want to exploit "eco" ontologies for semantic annotation of services to do "smart discovery and integration"
• Our solutions are meant to be generic, and applicable for KEPLER

Ontology Languages

• An ontology is a set of concept (class) names, subconcept (subclass) links, named (directed, binary) relationships between concepts, and constraints (cardinality, equivalence, conjunction, disjunction, etc.)
• In SEEK, we've adopted the Web Ontology Language (OWL)

Semantic Annotationa

• A semantic annotation assigns a 'selected item' to an ontology expression.

• Items
• Datasets: An entire dataset or some portion (a single table, one or more attributes, one or more data values, etc.)
• Workflows and components: A workflow, a workflow component, or some portion (parameters, ports, substructures of a port type, etc.).

• Item Selection
• Expressed in general as queries, however, simpler expressions can be used, e.g., XPath/XPointer addresses or possibly using EML identifiers

• Ontology Expressions
• An ontology expression defines the semantic "context" of the item selected
• Ontology expressions can be as complex as SQL-style update queries or simpler, e.g., as single concepts or paths in an ontology
• Path Example: Measurement.spatialContext.loc.latDeg specifies the location of a Measurement's spatialContext as a latitude in degrees

Architecture

• Repositories
• Ontology(ies)
• Datasets (or metadata stating how to obtain the datasets)
• Workflows and Workflow Components (or metadata, etc.)
• Semantic Annotations

• "Smart discovery and integration" needs access to each of these components:
• For example, to search for a workflow component, the sms engine would search through semantic annotations, and when an annotation matches, obtain the corresponding component.
• You might also want to organize (for browsing) all actors according to their annotations; so need to iterate over actors, or similarly, for datasets.

• SMS-Based Applications
1. Browsing/Keyword Search
2. Workflow and/or dataset browsing and keyword-search based on ontology terms
3. Find "compatible" actors
• Given an actor, find semantically compatible actors that can be connected as an input or output to the actor (this might result in a dataset as input, e.g.)
• Requires port inputs/outputs to be semantically annotated
4. Workflow "analysis"
• Given a workflow, check that each connection (input/output) is semantically compatible.
• As part of analysis, annotation propagation ...
5. Workflow-component structural integration
• Given two components that are semantically compatible, determine a structural transformation (either another component or a transformation step) to make them structurally compatible.
• May be a place where SCIA can contribute, to derive structural transformations.
6. Dataset merging/integration
• Search for "similar" datasets (that could be potentially "merged" or integrated)
• Define a dataset of interest (via an ontology-style query), find/combine datasets as integrated "view".
• Perhaps a place for SCIA to contribute?

• Tools

*** Ontology Editors/Browsers

*** Semantic Annotation Editor

*** Ontology-based query rewriting/answering

• "Smart" Actor Search in Kepler

A very simple keyword-based search implementation within Kepler.

Fakes out: workflow component LSIDs, an actor repository (as a ptolemy xml config file), annotation repository (xml file), LSID service.

The "ontology" is a simple hierarchy. No rels, etc.

• What's Needed for KEPLER

** Ontologies

There basically aren't any.

There also aren't any tools. No tools within Kepler.

** Repositories ...

The Obj. Mngr. can help! We don't have repositories for workflows/components, ontologies, annotations, or datasets in KEPLER.

For annotations, need a searchable "index" of annotations and ids (for components, datasets, etc.), and a mechanism to "retrieve" those items.

For performance, I wonder though if the "index" should be in memory.

** Semantic Annotation Editor This doesn't exist either ... lots of ways/approaches here.

Need GUI design for this. Also, need a good way to access/browse a component/dataset and its attributes, such as is ports and their input/output types.

Similar with datasets

The challenges are making this tool easy to use, and accessible within Kepler.

** Basic Kepler Interfaces / GUI Design

Like for searching, checking semantic compatibility (can steal unit resolver), explanation of semantics (like for searching, etc.)