Ontologies as an Organizational Aid

Ontologies can help organize information about a topic domain, either within a data system or across different data systems. They can achieve this because of the basic features of ontologies. These features:

• Provide a means for defining concepts so that they can be consistently accessed by a computer.
• For example, MMI is developing a platforms ontology, which defines different types of oceanographic platforms and their properties (e.g., mobile or not), so that searches for data sets by platform type, or by properties of a platform, can be done automatically across data systems, regardless of the specific terminology used.
• Provide hierarchical frameworks for organizing concepts.
• Classes, subclasses, and “is a part of” are examples of the hierarchical organizational framework provided by ontologies for concepts. In the MMI platforms ontology, the superclass concept is "Platform", with its subclass concepts, "AirAndOuterSpaceBasedPlatform", "EarthBasedPlatform", and "WaterBasedPlatform".
• Permit the articulation and accessibility of relationships between concepts (instances) and properties.
• These relationships can then be used automatically by computers to infer additional associations between concepts. Ontologies can automatically classify concepts according to the properties defined for those concepts. For example, a water-based sensor could be classified as a subclass of a sensor, simply if a property of the water-based sensor is also a property of a sensor. In other words, the property can be used to classify water-based sensors as a type of sensor, without having to manually create water-based sensors as a subclass of sensors.
• Can be extended to provide any kind of relationship, or mapping, between individual terms in separate ontologies.
• For example, at the MMI workshop, Advancing Domain Vocabularies, a sensors working group identified and mapped sensor-related terms from several vocabularies, including those from WHOI, MBARI, LDEO, SIO, TAMU, NGDC, CO-OPS, ACT, and BODC. This work was the precursor to the development of the MMI Sensor Ontology project, the goal of which is to develop a sensor ontology, based on existing vocabularies and the mappings initiated at the MMI workshop.

Through application of these features, ontologies provide a variety of higher level knowledge-management capabilities, each of which helps organize information and knowledge. Ontology applications:

• Provide consistency to the terms used in metadata records.
• Consistency in metadata is essential to keeping information organized, making it discoverable, and enables interoperability between data systems.
• Can be used to generate knowledge bases about one or more specific domains(s).
• Each ontology represents a set of knowledge about some topic area. By connecting related ontologies, the knowledge framework can be extended to cover a wider domain.
• Provide more powerful terms for filling out metadata records, so that they can better represent information.
• By formally defining the terms used in metadata records, and enabling those terms to be mapped to other terms relevant to that community, ontologies extend the completeness, precision, computability, and extensibility of the metadata records. (Each term in an ontology can carry with it the context of the entire model, due to more complex semantic statements and the inference capabilities of Description Logics). For example, ontologies can define the values used to complete metadata fields like Keywords. The terms can then be mapped to other vocabularies, and interoperability is facilitated between different metadata records, regardless of the specific terminology used.
• Provide greater descriptive detail for metadata models and metadata specifications, so that they can more clearly capture and organize information.
• By formally defining the field names used in metadata standards, and enabling those terms to be mapped to other terms in other standards, ontologies create a more precise and interoperable standards framework. For example, ontologies can organize and characterize the terms in different metadata standards (e.g., "creator" vs. "author"), enabling both better understanding by people using the standard, and greater interoperability between different standards.
• Support discovery and understanding through interactive navigation.
• The relationships captured by ontologies are analogous to those in topic maps: they tie together the different terms of the ontology. Visual presentation of these relationships provides a different way to view the knowledge model represented by the ontology, and interactive tools allow it to be easily explored, often with serendipitous results. These explorations can be built in to the interfaces used to discover data sets and other scientific materials, allowing searches to be qualified in ways that make sense for the particular subject domain.

Have a specific question about ontologies? Ask MMI!