Upper Ontology Design for ApplicationBased Spatial Ontologies Eric

Upper Ontology Design for Application-Based Spatial Ontologies Eric Little, Ph. D D’Youville College National Center for Ontology Research (NCOR) National Center for Multisource Information Fusion (NCMIF) Buffalo, NY USA little@dyc. edu eglittle@eng. buffalo. edu

The Structure of an Ontology • Upper-Level (Formal): – Most general categories of existence (e. g. , existent item, spatial region, dependent part). – This Level of the ontology is rationally driven, meaning it is the product of philosophical reasoning. – Relies on a sound metaphysical description of the world (e. g. , realism).

The Structure of an Ontology • Domain-Specific Level – Contains categories that are specific to a particular domain of interest (disaster, military/defense, medicine). – This level of the ontology is empirically driven, meaning it is produced by gathering expert knowledge about a given domain of interest. – The expert knowledge is used to create a consistent and comprehensive lexicon of terms.

Synthesized Ontology Model Formal Ontology (Rational) Application-Based Formal Ontology Collection of Non-Formal Information Needs (Empirical)

Ontologies vs. Taxonomies Urban Environment Taxonomy A IED Taxonomy B ONTOLOGY Dirty Bomb Taxonomy C ETC…

Using Knowledge Representation & Reasoning (KRR) to Conjoin Taxonomies SNAP Taxonomy (Spatial Items) Transcategorical Relations Represented in KRR Example: An Intentional Act is a Psychological Act that depends on an agent to instantiate it. It stands in a relation of dependence to other items such as neuro-biological states. SPAN Taxonomy (Temporal Items)

Relating Ontology to Other Engineering Practices • Ontologies inform the design of other engineering systems (e. g. , agent-based sys, decision support sys, predictive analytics, etc. ) by providing a structured comprehensive picture of their domains. – Many engineering practices require a more principled basis for their design. • Engineering systems constrain the ontology by providing inputs such as: – User needs – Domain specificity – Computational tractability • If you give philosophers carte blanche, remember … fools and their $ are easily parted.

Higher Level Fusion The purpose of higher level fusion is to develop probable explanations of a situation based on prior knowledge and incoming transient information to produce a coherent composite picture of the current situation along with a prediction of consequences. A dynamic situational picture is the result of reasoning about objects, attributes, aggregates, relationships and their behavior over time within a specific context. The process of building the dynamic situational picture requires formally structured and computationally tractable domain representation.

What kinds of ontologies are needed for High-level fusion & STA? • Low – level fusion can be done (to a large degree) using existing tools such as OWL, Protégé, DAML - Oil, etc. • However, higher-level fusion processing is concerned with providing comprehensive and consistent descriptions of highly complex world states. • Hence we need a more “industrial strength” (cf. Musen) approach than is provided by current fusion ontologies.

Relations Between Situational Objects at Different Levels of Granularity • • Inter - Relationships: 1) Relationships between situational items of different types. 2) Relationships between items and aggregates of items of a different type. 3) relationships between aggregates of objects of different types Intra - Relationships: 1) Relationships between different physical objects or their respective attributes/properties. 2) Relationships between different clusters/aggregates of objects in the same group. Physical objects – Physical objects (PO-PO) PO- Aggregates of PO Aggregates Interclass Intraclass Combinations of ES (CES) – Combinations of ES Relations Temporal Spatial Elementary Situation -Elementary situation (ES-ES) CES-ES Processes-Processes (process aggregation) Events-Events (event aggregation)

Ontologize this… Frank White (Workshop II on Ontologies and Higher-lvl Fusion – Beaver Hollow

Existing fusion ontology models often confuse various kinds of relations Temporal Relations Spatial Relations Situation Awareness (SAW) Ontology Model for Battlefield Relations (C. J. Matheus, M. M. Kokar, and K. Baclawski. (2003)

It gets worse… Complex Relation Type

Examples of Important Relationships SNAP relations Topology/ mereology Disjoint Joint Overlap Cover Reachable Unreachable Contain A part of Direction Along Towards East West South North Similar Opposite Distance SPAN relations Size Far Small (er) Very far Large(r) Near Same Very near Relationships between time points Relationships between time intervals Before, At the same time, Start, Finish, Soon, Very soon, Resulting in, Initiating Disjoint, Joint, Overlap, Inside, Equal Disaster Examples: “Close to a hospital” “Cluster A is larger than before” “Along the wind direction” “Distance between Clusters A and B is smaller than before” “Casualty cluster A overlaps with building cluster C”

Building Reasoning Processes with Ontologies SNAP Ontology FUSION SPAN Ontology -Spatial Items Of Interest Reasoning about relations represented in Ontology -Temporal Items Of Interest Transcategorical Ontology (Objects + Processes)

Segment of SNAP Kharkiv Nuclear Facility Ontology

Segment of SPAN ontology for Kharkiv Nuclear Facility

Small Representative Sample of the SNAP Dis-Re. O Ontology w/ CWA Bisantz, A. , Rogova, G. , Little, E. (2004) “On the Integration of Cognitive Work Analysis within a Multisource Information Fusion Development Methodology, ” Proceedings of the Human Factors and Ergonomics Society Annual Meeting, New Orleans