The Need for Ontologies in Natural Resource Research and Management

From forestDSS
Jump to: navigation, search

( from FSWiki-a Wiki for the Forest Service community - MJ Furniss 2007)

What would a map look like that showed the different concepts and theories underlying the sciences involved forest ecology? How would we even begin to assign spatial coordinates to the myriad abstract ideas this involves? Let's break it down. Every discipline has its own unique vocabulary, such as the language that is used for discourse within a particular realm of science. Some of that language already adheres to formal structures that we agree to.Taxonomic reference to species, in particular, adheres to the Linnaean hierarchy of domain or kingdom, phylum, class, order, family, genus and species. Kingdom - Phylum - Class - Order - Family - Genus - Species –

You may remember the handy mnemonic for this hierarchy: King - Phillip - Came - Over - For - Great - Spaghetti –

In other language used in scientific disciplines, the hierarchy and Relationship between terms is fuzzier. There are several thesauri (thesauruses?) that attempt to list descriptive terms in agriculture and related sciences with cross-references, mainly to facilitate storage and retrieval of documents.

But a thesaurus doesn't go much further than listing terms, their synonyms, related terms and whether a term is broader or narrower than a related term. 

                PHOTOSYNTHESIS
           UF+ carbon assimilation 
         UF+ carbon dioxide fixation 
                    / | \
                   /  |  \ 
BT energy metabolism  | NT calvin cycle 
       BT metabolism  | NT cam pathway 
             -        |        - 
           RT carbon dioxide enrichment 
               RT gas exchange 
             RT leaf area index 
             RT photoreceptors 
        RT photorespiratory pathway 
               RT plastids 
          RT primary productivity 
         RT source sink relations 
                     -
LEGEND
 BT broader term
 NT narrower term
 RT related term
 USE use
 uf used for
 uf+ used for in combination
 (http://www.fao.org/agrovoc)

When we attempt to describe scientific theory, we need much richer descriptions. We must include terms that represent objects or concepts, and alternate choices of terms, as in a thesaurus. We must also include the possible attributes of each term, so we can fully describe it in a standard way. Then we must describe the relationships between terms, so we get a clear picture of how they fit together. Is the relationship causal, one object acting upon another? Do the objects interact with one another? Can we describe the action or reaction as a function? Maybe we have more than one theory, thus more than one possible function to ascribe to a relationship?

Information science, particularly that realm we call "artificial intelligence," grapples with these issues. If we could have a way of organizing and clearly representing all of the concepts and relationships in a scientific discipline, we would have a unique kind of map. We could navigate this map following many trails. A specification of a representational vocabulary for a shared domain of discourse -- definitions of classes, relations, functions, and other objects in that domain -- is called an ontology. An ontology may be represented like so:

PHOTOSYNTHESIS... is an instance of energy metabolism metabolism ...depends upon calvin cycle cam pathway leaf area photoreceptors photorespiratory pathway plastids primary productivity ...depends upon source sink relations ...produces carbon dioxide and so on

Your school of thought might see it differently, in which case we could present ontologies showing your view and my view, each as different "domains" describing the same concept,photosynthesis. We could juxtapose our two views, see where they are similar and where they are different. We might find agreement on areas where further detail is needed to more fully understand the concept. Ontologies have the potential to be used as accurate maps of concepts, Domains of thought, and whole realms of science. They can provide the underpinning root structure that may be traversed along pathways that quickly lead to more and more focus on a specific area of interest, since each decision about navigating an ontology should "zoom" by a factor of two, dividing the remaining field of view in half. Contrast this to the relatively random walk that now characterizes internet search engines, and you'll start to imagine what's coming. A search for "photosynthesis" at google.com now produces a remarkable 6,980,000 results in less than three tenths of a second. I'm still working on scrolling through the results.

Further reading:

Retrieved from "http://www.forestdss.org/wiki/index.php?title=The_Need_for_Ontologies_in_Natural_Resource_Research_and_Management&oldid=11062"