Upper-Level Ontology Considerations for the Geospatial Ontology Community of Practice Eric Little, PhD D’Youville College Center for Ontology & Interdisciplinary Studies National Center for Ontology Research (NCOR)
Purpose of the GSOCoP all information about government assets and services - and about the citizen beneficiaries of those services - has a location component. (E.g.,) education, voting, grant eligibility, benefit claims, medical care, recreational opportunities and more. Governments need to pay attention to changes in land use, land cover, shorelines, population, wealth distribution, transportation, hazards, threats, energy systems, etc. (From: S. Bacharach, 2005)
Theoretical and Implementational Concerns Information will come from different sources – thus the need for an overarching architecture (Reference Ontology). –How to provide this? What is the “ontological pedigree” needed? How does the community build a consensus (if there is one)? Implementational concerns (robustness vs. computational tractability, user needs, legacy data systems)
Theoretical Concerns (cont.) Differences in topographical representations –cartographers, information scientists, geomorphologists and environmental modelers (field of elevations or as some discrete approximation to such a field). –Pilots, explorers,anthropologists, ecologists, hikers, and archeologists identify (special sorts of) objects (hills and valleys, mountains and plains, barrows and trenches), with locations, shapes, and often names of their own (Mark & Smith, 2003) –Need for REALISM + PERSPECTIVALISM Problems of Vagueness and Granularity (e.g., boundaries, dependent parts, etc.) Need a good theory of RELATIONS
Geographical Entities are Indeterminate and Ambiguous Objects Philosophically speaking: Where does the mountain begin and the valley end? How can we derive a common semantics which can refer accurately to these kinds of objects?
Ambiguity/Vagueness The problems of ambiguity, vagueness, granularity, etc., point to the need for metaphysically-based upper ontologies. In other words, we need to be analytically consistent about many kinds of geographic items. We also need to capture issues surrounding conceptualization of these items. –Helps with CSE, WDA, and other kinds of user-centric needs.
How Can Upper Ontologies Help? Upper Ontologies are DOMAIN INDEPENDENT, so they allow for disparate systems to reason about basic elements common to all ontologies (processes, objects, spaces, etc.) Can be used to TRANSLATE across different domain ontologies by providing appropriate (inter-theoretical) semantic content. Conceptual mapping becomes easier and more consistent, if there is a good upper-ontology being utilized.
Many Choices Exist… Varieties of Upper Ontology Tools include (but are certainly not limited to): –DOLCE –SUMO –WonderWeb –OCHRE –BFO –Upper CYC –Information Flow Framework (IFF) […] OBO Foundry
Taxonomies
Taxonomies vs. Ontologies
Relations are Key
Trans-Ontological Relations (from BFO) Provide complex relations which exist BETWEEN the orthogonal ontologies of SNAP (spatial items) and SPAN (temporal items). Can be used to model many kinds of relations between endurants and perdurants –(e.g., an agent’s role as doctor and husband, given certain contexts (contracts/promises) and relationships to various social organizations (medical schools, legal systems).
SNAP-SPAN Relations SNAP Independent Items Dependent Items Spatial Regions SPAN Processural Entities Processural Events (instants) Temporal Regions Relations can be of the general types: (SNAP SNAP) (SNAP SPAN) (SPAN SNAP) (SPAN SPAN)
Formal Relation Types (SNAP-SNAP & SPAN-SPAN) SNAP-SNAP –Genidentity. (A=a+b & A=A) –Transgranular Part- Whole Relations. (P.O. – Aggregate) –Subsumtion Relation SPAN-SPAN –Genidentity. (A=a+b & A=A) –Transgranular Part- Whole Relations. (Event/Process – Aggregate) –Subsumtion Relation
Most Basic Formal Relation Type (SNAP-SPAN) Segmentation (Individuation) –Tennis Match (fiat boundary of sets, games) –Occipital Lobe (fiat boundary determined by function of cells) Segmentation in SNAP can be bona fide or fiat (distinct substantial boundaries) Segmentation in SPAN is fiat (processes exist as continua only – everything is in flux)
SNAP Indep. SPAN Participation (subset of Dep. Relation) (substance affects process) –Perpetration (Active Agency) Initiation (=activation, begin, commencement) Perpetuation (=continuation, sustainment) Termination (=cessation, end) –Influence Facilitation (influence a PART of process/event) Hindrance (impede a PART of process/event) –Mediation (Indirect influence on a process) –Patiency (Passive Agency) (A process carried out in a non- agentive manner by a substance – e.g. digestion)
SNAP dep. SPAN Realization (Dep. Substances are realized via their behaviors/activities) –Initiation –Persistance –Termination Includes, Roles, Powers, Functions, Liabilities, Perspectives, etc., which may exist even when not being immediately realized (when one is asleep, when an object is not currently in use)
SPAN SNAP (Indep. & Dep.) Involvement (converse of Participation) –Creation –Sustainment of Being –Degradation –Destruction –Affection (SPAN SNAP dep.) Creation Continuation Degradation Destruction –Demarcation –Blurring
SNAP Spatial Region SPAN Temporal Region Projection –Process PROJECTS onto a temporal axis (SPAN SPAN) –Substance PROJECTS onto a period of time (SNAP SPAN) –Process PROJECTS onto a spatial location (SPAN SNAP)