[seek-kr] next steps for KR/SMS

[Rich Williams]

Great job summarizing our conversation Mark!

Here are some thoughts on what an “ontology of niche models” might look like
that I was writing when Mark and I talked.  It is somewhat repetitive with
the first part of Mark's email but might be useful.  Also be warned that it
is based on a very rudimentary understanding of the models themselves.  Now
I'm offline until Monday!

Rich

----------------------------------------------------------------------------
---

I see two very different levels of description that could be done for the
ecological niche model ontology.  There is of course overlap, but I think
the division is useful.

1)	Description of inputs and outputs of GARP (or other niche models).  This
would include concepts like: species, environmental conditions/properties,
geographic location and range, geographic distribution of species existence
and abundance, geographic distribution of environmental properties.

This kind of ontology would A) facilitate the discovery of data to feed into
the model under consideration.  B) allow description of the output of the
model, and so facilitate the formulation of complex data analysis pipelines.

2)	How ecological niche models, and GARP in particular, fit into the world
of ecological and environmental models and data analysis tools.  Concepts
might include: species, environmental properties, niche, niche dimension,
species distribution/range, climate change.

Uses of a description at this level might include A)  a scientist
researching the effects of climate change, who would discover a set of
algorithms that would address the effects of climate change on species
distributions.  B) a scientist researching species distributions in a
particular region.  C) a scientist researching what environmental properties
are important in determining the niche of a particular species

The important point of this is that (2) involves many more high level
concepts from ecology and environmental science, and enables a very
different set of functionalities for researchers.  I think that creating
this higher-level ontology will be considerably more challenging than the
data-description ontology needed to accomplish (1) (not that (1) will be an
easy task!).  While it is important for SEEK to address both levels of
description, I would advocate initially concentrating on the data
description and discovery issues raised by (1), while keeping (2) as a
longer range goal.

-----Original Message-----
From: seek-kr-admin at ecoinformatics.org
[mailto:seek-kr-admin at ecoinformatics.org]On Behalf Of Mark Schildhauer
Sent: Wednesday, July 16, 2003 12:55 PM
To: Deana Pennington; seek-kr at ecoinformatics.org
Cc: Rich Williams; Bertram Ludaescher
Subject: [seek-kr] next steps for KR/SMS


Hi Deanna and KR folks,

Here's a shot at a summary of some recent discussion among Bertram, Rich
and myself relative to what we might be able to accomplish in the
short-term relative to KR/SMS, and specifically --what Deanna could
possibly do at the upcoming meeting with the BEAM ENM (ecological niche
modeling) group in a couple of weeks, and what we can present at the
AMS/KR tutorial/workshop in Albuquerque at the end of next month.
Comments welcomed!

For upcoming BEAM meeting in Mexico--
1) It will indeed be valuable (as Deanna suggested) to identify and
clarify common terms and concepts used by ENM scientists.  This can be
done in two distinct areas--
    a) clarifying the specific types of data that go into parameterizing
ENM models, and their outputs--relative to themes, attributes, formats,
sources, etc.  This will be particularly relevant for parameter
ontologies that inform the AMS.  Should specifically think about how to
describe data needs in general terms sufficient to clarify how
*additional* data sources might be incorporated or assimilated by the
models.
    b) identifying higher-level conceptual issues related to reifying
"ecological niches"-- formal definitions, algorithms, etc.  This will be
relevant not only for heuristic value to scientists (if we can find good
ways to visualize and reconcile these belief/concept maps), but also for
enhanced querying (e.g., via term expansion) on EcoGRID, and what
Bertram calls "concept-based indices" for identifying tools within an
AMS library.

Ideally we'd want to capture these thoughts in a formal framework, e.g.
using an RDF or OWL plugin to some tool like Protege.  Unfortunately,
based on Rich's experience, while Protege is an excellent tool, it is
NOT suitable for a casual end-user.  A current best-fit model would
involve a "knowledge engineer" working closely with domain scientists to
clarify and formalize an ontology.  But luckily, we feel that a
whiteboard style capturing of directed graphs, lists of terms with
definitions, etc. can be accomplished at the BEAM meeting, and later
formalized through discussion, e.g., between Deanna and Rich.  So,
Visio, Illustrator, graphVIZ-- whatever-- can be used to capture
concepts, relationships, definitions, etc.  And we can hopefully
formalize these later through dialogue among e.g., Deanna, and Rich.

2) Bertram is going to attempt to get out (within the next several days)
a working copy of his Prolog application that could do simple unit
conversions (based on the EML unit dictionary) using an ontology and
inferencing approach.  This would be a simple "proof of concept" of the
utility of KR/SMS technology.  We thought this might be useful for
Deanna to carry down to Mexico, that displays some of the potential
utility of  this approach.  Deanna-- when is the latest that you might
be able to receive/download some application prior to departing for Mexico?

3) Rich is actively thinking about niche theory, and hopes to begin work
on some ontology creation for this subject area soon.  He has recently
checked in to the SEEK CVS repository a Protege-compliant ontology for
Food Webs, that could serve as good exemplar for ecologists wanting to
build ontologies.  Rich feels that there are some good visualization
capabilities for ontologies within Protege using the graphVIZ or
touchGraph plug-ins, so you might want to give these a try.

For AMS/KR workshop at end of August--
We are still thinking Protege is a good tool for working with
ontologies.  Very strong and diverse developer community, lots of
momentum, mature code, etc.  Need to work on simplifying it (according
to Rich, can highly customize the appearance of Protege for different
audiences), testing plug-ins for visualization, choice of OWL, RDF, etc.

Have a couple of example ontologies (Food Web, Ecological Niche) in hand
to walk-thru with scientists

Have working application for unit conversions (Bertram's prolog app).
 Might want to investigate other Protege plug-ins for this purpose,
including Algernon.

Some of these tasks can be moved forward by Rich, and Sean, Bertram's
new full-time SMS programmer, who started work yesterday!

Need some work on How-To: create/edit ontologies, visualize these, reason.

Bertram and I are going to be on vacation for most of August.  Rich is
getting married this weekend (congratulations!) and his life is hectic
this week, but settling into blissful tranquility the week after.  Sean
is jumping onboard this week, and will be able to actively pursue some
of these issues (right, Bertram?) along with Rich.  Chad is also working
on the AMS, and will be keeping in mind linkages with the KR/SMS component.

Finally, I am attaching a preprint that Deanna received from Mark
Gahegan during her recent visit with the Geovista project.  This is an
interesting document with a lot of apparent overlap with our goals and
interests in KR/SMS.  Mark Gahegan said we were welcome to share this
with the SEEK KR participants.

Any comments, corrections or additions to this note are welcomed!

Cheers,
Mark

--
Mark P. Schildhauer, Ph.D. --  Director of Computing
NCEAS --  National Center for Ecological Analysis and Synthesis
735 State St., Suite 300       Santa Barbara, CA   93101-3351
Email: schild at nceas.ucsb.edu   WEB: http://www.nceas.ucsb.edu
Phone: 805-892-2509            FAX: 805-892-2510