Previous Page Table of Contents Next Page


2 Methodologies to support ontology integration


2.1 Heterogeneous systems give heterogenous interpretations

An example of how formal ontologies can be relevant for fishery information services is shown by the information that someone could get if interested in aquaculture.

In fact, beyond simple keyword-based searching, searches based on tagged content or sophisticated natural-language techniques require some conceptual structuring of the linguistic content of texts. The four systems concerned by this project provide this structure in very different ways and with different conceptual 'textures'. For example, the AGROVOC and ASFA thesauri put aquaculture in the context of different thesaurus hierarchies; an excerpt of the AGROVOC result is (with a penchant for kinds of techniques and species):

AQUACULTURE

uf aquiculture
uf mariculture
uf sea ranching
NT1 fish culture
NT2 fish feeding
NT1 frog culture
...
rt agripisciculture
rt aquaculture equipment
...
Fr aquaculture
Es acuicultura

while the ASFA result is substantially different (it seems to stress the environment for aquaculture):

AQUACULTURE
uf Aquaculture industry
uf Aquatic agriculture
uf Aquiculture
NT Brackishwater aquaculture
NT Freshwater aquaculture
NT Marine aquaculture
rt Aquaculture development
rt Aquaculture economics
rt Aquaculture engineering
rt Aquaculture facilities
...

FIGIS reference tables may interpret aquaculture in still another context (taxonomical species):

Biological entity
Taxonomic entity
Major group
Order
Family
Genus
Species
Capture species (filter)
Aquaculture species (filter)
Production species (filter)
Tuna atlas spec

and oneFish directory returns the following context (related to economics and planning):

SUBJECT
Aquaculture
Aquaculture development
Aquaculture economics @
Aquaculture planning

With such different interpretations of aquaculture, we can reasonably expect different search and indexing results. Nevertheless, our approach to information integration and ontology building is not that of creating a homogeneous system in the sense of a reduced freedom of interpretation, but in the sense of navigating alternative interpretations, querying alternative systems, and conceiving alternative contexts of use.

To do this, we require a comprehensive set of ontologies that are designed in a way that admits the existence of many possible pathways among concepts under a common conceptual framework. This framework should be domain-independent, flexible enough, and focused on the main reasoning schemas for the domain at hand.

For example, the domain-independent ('upper') ontologies should characterise all the general notions needed to talk about economics, biological species, fish production techniques; while the so-called core ontologies should characterise the main conceptual habits (schemas) that fishery people actually use, namely that certain plans govern certain activities involving certain devices applied to the capturing or production of a certain fish kind in certain areas of water regions, etc.

Upper and core ontologies [7,8] provide the framework to integrate in a meaningful way different views on the same domain, such as those represented by the queries that can be done to an information system.

2.2 Methods applied to develop the integrated fishery ontology

Once said that different fishery information systems provide different views on the domain, we directly enter the paradigm of ontology integration, namely the integration of schemas that are arbitrary logical theories, and hence can have multiple models [9]. As a matter of fact, the thesauri, topic trees and reference tables used in the systems to be integrated could be considered as informal schemas conceived to query semi-formal or informal data bases such as texts and tagged documents.

In order to benefit from the ontology integration framework, we must transform informal schemas into formal ones. In other words, thesauri and other terminology management resources must be transformed into (formal) ontologies.

To perform this task, we apply the techniques of three methodologies: OntoClean [8], ONIONS [10], and OnTopic [11]. The first deals with the use of upper ontologies and general principles for core and domain ontology building, the second describes several methods for enhancing the informal data of terminological resources to the status of formal ontology data types, the third shows how to create links between topic hierarchies and ontologies.

In Figure 1 a class diagram is shown of the informal and formal data types taken into account, while in Figure 2 a state diagram is sketched of the methodology used to extract and refine the informal data from the fishery information systems.

In the next section we briefly describe:


Previous Page Top of Page Next Page