|
|||
|
This is version 53.
It is not the current version, and thus it cannot be edited. 1 Introduction
2 The Eco-Ontology Core
This section describes the fundamental building blocks of the eco-ontology -- called the eco-ontology core (or just the core). The core is represented as a single package, and all other ontologies defined within (or that extend) the eco-ontology are built from the core. The purpose of the core is to establish the set of basic semantic constructs that can be used to define more complex ecological concepts and relationships.
Things
Every concept in the eco-ontology extends the generic concept EcoOntThing. The name "thing" is often informally used in ontologies to denote the set of all things. We use the term object instead of "thing", but we retain the convention of using "thing" to name generic concepts. Figure 1 defines EcoOntThing. As shown, objects can be named using the hasName attribute. Object names are optional, denoted by the '0' in the cardinality restriction '0..*'; and an object may have more than one name, denoted by the '*' in the cardinality restriction '0..*'.
Parts and Wholes
Objects are often described in terms of their structural relationships. A basic structural relationship is the connection between parts and wholes in which one object serves as a composite, containing zero or more other objects serving as parts. Figure 2 shows how parts and wholes are defined in the core. The concepts Composite and Part are sub-concepts of EcoOntThing, and are related through the hasPart and partOf' relationships.
The hasPart relationship relates a composite object to zero or more parts. We denote a hasPart relationship between a composite c and a part p as hasPart(c, p). The partOf relationship relates a part object to its associated composite object, that is, partOf(p, c). A part object can participate in zero or more composite objects -- thus, parts can be shared by composites.
The hasPart and partOf relationships are inverse and transitive. The inverse relationship states that for every composite object c and part p, hasPart(c, p) if and only if partOf(p, c). The transitive relationship states that for every composite object c and parts p1 and p2, if hasPart(c, p1) and hasPart(p1, p2) then hasPart(c, p2). The inverse relationship also implies that if partOf(p1, c) and partOf(p2, p1) then partOf(p2, c).
There are four ways that part-whole relationships are specialized in the core, shown in Figures 3 - 6. The first specialization is shown in Figure 3 and defines inseparable parts, which are parts that are always contained within at least one composite -- that is, the part object does not exist separately from a collection. Inseparable parts are represented using the inseparablePartOf relationship and the concept InseparablePart. As shown, the inseparablePartOf relationship is a sub-relationship (that is, specializes) the more general partOf relationship.
The second specialization is shown in Figure 4 and defines essential parts, which are parts that a composited requires. In general, composite objects can contain a variety part types, for example, a car is made up of wheels, a chassis, bumpers, and so on. For a car, wheels are generally considered essential parts, however, sunroofs are not. Essential parts are represented using the hasEssentialPart relationship and a the concept EssentialPart.
The third specialization, shown in Figure 5, describes unique parts. A collection object will have at most one unique part object, if a unique part is defined for the collection. A chassis is a unqiue part for a car, that is, any one car can have at most one chassis. Note that a chassis is also an essential part for a car. We permit part-whole specializations to be freely mixed.
The last specialization, shown in Figure 6, defines unshareable parts. An unshareable part object can be a part of at most one collection at a time.
Observations and MeasurementsLike parts and wholes, measurements (as well as the more general notion of observations) are fundamental building blocks in the eco-ontology. As shown in Figure 8, we define MeasurementConcept to highlight the special status of measurements in the core. In particular, MeasurementConcept is defined as an abstract concept (denoted by the italics type in the figure) that all other measurement definitions fall under.
Observations (and therefore, measurements, which we describe further below) can be grouped into collections through the concept ObservationCollection. Observation collections can also be grouped into other collections. We use the abstract concept ObservationCollectionPart to denote the objects that can act as parts in an observation collection object. As shown in Figure 8, we use the part-whole constructs to describe these relationships between observations and observation collections. Note that the cardinality restrictions are not shown because they are implied by the part-whole constructs used. An observation is a generalization of a measurement. Informally, we define an observation as an assertion of the existence of an entity or of the existence of some property of an entity. A measurement goes further by asserting a value for a property. To illustrate, consider the statement "I saw a tree frog". This statement refers to an observation, in which the existence of a tree frog was observed. The statement "I noticed the tree frog had spots" is also an observation in which the existence of the property "spots" was observed for a tree frog. We say that for this case, the observation asserts there is some value for the property "spots" that was observed, but the value was not provided. Finally, the statement "I noticed the tree frog had ten spots" is a measurement, in which the existence of the value ten is being asserted for the "spots" property of a tree frog.
|
This material is based upon work supported by the National Science Foundation under award 0225676. Any opinions, findings and conclusions or recomendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF). Copyright 2004 Partnership for Biodiversity Informatics, University of New Mexico, The Regents of the University of California, and University of Kansas |