By	ElizabethAuden
	AnitaRichards
	TonyLinde
Goal	1. Produce subset of astronomy ontology 2. Produce registry prototype using ontology
Technology	JBoss: EJB server
	Jena: Java api for reading DAML+OIL docs
	OilEd: editor for producing DAML+OIL docs

Also check out the OntologyReading^? page for background reading material.

-- TonyLinde - 04 Jul 2002

---

Interim Write-up (20020816)

This will not be a long document, for several reasons. Firstly, most of the work we've done has been written up already (see OntologyDocs). Secondly, the demo is far from finished (which explains the heading) but, given that it was supposed to produce some results by now, this report will document what we have achieved. Thirdly, the learning curve for ontology work is steeper than anticipated.

The driver for this work was the belief that the UCD structure produced by CDS was insufficient as a description of astronomical concepts and work to serve as a basis for a VO registry of resources. We had heard about work being done in the bioinformatics field on devising ontologies and using them to describe and access widely dispersed resources and thought it worth investigating whether a similar approach would work for astronomy.

We began work at the beginning of July (OntologyFirstTry), producing a list of 'big' concepts and looking at topics which are subsumed within those concepts. This exercise seemed to produce more questions than answers. The diagram produced looked a bit like a prettier version of the UCD structure. The one divergence was the introduction of multiple inheritance (not that we called it that at the time): 'Spectral Energy Distribution' is subsumed both by the 'Wavelength' and 'Energy' concepts.

A second approach was started by Elizabeth and Anita (OntologyUserData) to collect, under a number of headings, the types of information that astronomers deal with in their work. This is an important contribution as it shows the concepts which different fields have in common and those in which a field might be unique. It is these terms which any future AstroOntology will have to accomodate.

Elizabeth had also been making progress in using and understanding OilEd, an ontology editor which works using DAML+OIL constructs, and the Jena Toolkit, a Java api for manipulating RDF models.

The next significant advance came after Anita and Tony attended a tutorial in Ontology run by Carole Goble and Nigel Shadbolt as part of GGF5 (see 'Presentation' link at http://www.semanticgrid.org/). An ontology can be seen as a collection of taxonomies, and each taxonomy represents one relationship type. So, one relationship type might be 'is made up of' and you could have a taxonomy whereby the universe 'is made up of' clusters of galaxies, galaxies, dark matter etc; a cluster of galaxies 'is made up of' galaxies, dark matter, etc. (My astronomy knowledge runs out here) Another taxonomy might be 'subdiscipline', so Astronomy has the subdisciplines, Observational Astro and Theoretical Astro; and Observational Astro has the subdisciplines XRay Astro, Radio Astro etc.

We met a couple of weeks ago to explore what might be Astronomy's OntologyRelationshipTypes in an interesting and dynamic meeting in Oxford. This list proved to be an excellent starting point. It generated more discussion amongst us relating to the 'right' direction for relationship types, what use could be made of namespaces, etc.

Following this meeting, Tony was able to produce a number of TrialOntologies. The first three attempted to represent some simple astronomical constructs but (as Anita has pointed out in a private email) they will need significant work before being able to represent any true state of astronomical knowledge.

The last trial was an attempt to import the UCD structure. This was successful and will be used in future trials to link generic concepts to the actual UCD structures. In this way it might be possible, in a trial and as a short term measure, to run a registry using concepts from an AstroOntology which then link into VizieR to retrieve catalogue information.

Finally, we had a meeting with a couple of people working on the bioinformatics ontologies at Manchester - OntologyMtgAtManchester20020815 - and learned a lot from them which we can put to use in future work in this area.

---

I would recommend that this work continue with investigations into:

• workflow and job control ontologies
• resource registry storage and access using ontological approaches

with an end date of 30-November. The goal should be to produce:

1. a first draft, skeletal, AstroOntology, incorporating UCD and VizieR information
2. a register of (a few) UK-based astronomical catalogues, each described in ontology-based terms
3. an ontology-based register access method
4. an ontology-based workflow, driving a register access web site

with priorities in the order above (in the expectation that all of the above will not be possible).

A subsidiary goal should be to maintain full involvement in the burgeoning semantic grid work and feed our results into that effort.

-- TonyLinde - 16 Aug 2002

---