An ontology is a system that contains terms, the definitions of those terms, and the specification of relationships among those terms. It can be thought of as an enhanced thesaurus - it provides all the basic relationships inherent in a thesaurus, as well as defining and enabling the creation of more formal, specific and powerful relationships. An ontology structures the knowledge in a domain and captures the meaning of concepts that are specific to that domain. The application of an ontology into an information management system provides end-users with tools that make indexing and therefore, retrieval, more precise.
Several satellite ontologies were developed within the broader context of the AOS project at FAO. As there is no standardized procedure yet on how to develop an ontology, different approaches were used in their construction.
Fisheries Ontology: The existence of several vocabularies and terminology tools in the areas of fisheries and aquaculture provided a good basis on which the ontology was built. Using ontologies for the linking and semantic mapping of concepts allows the applications to provide semantically accurate search results.[13]
Food safety ontology: The core ontology, which was built from scratch by subject experts, was extended using a toolset for ontology building. The first tool was a web crawler which identified domain related webpages to define additional relevant terms. A second method to extend the core ontology was to extract keywords from a sophisticated thesaurus on the basis of domain specific documents. To distinguish between domain-specific and non-domain-specific keywords the process was repeated with documents not relevant to the subject.[14]
In addition to traditional relationships, such as equivalence (equivalent terms), hierarchical (broader term/narrower term) and associative (related term), ontologies also allow for the definition of additional relationships, such as:
exact equivalence (synonym)
partial equivalence (words with similar meaning)
single to multiple equivalence
inexact equivalence (antonyms)
Defining non-traditional relationships offers the added possibility of distinguishing between concepts that are normally assumed to be similar in meaning. For example, very often "risk analysis" and "risk assessment" are used as synonyms. In a traditional relationship, they may be equated, whereas in a non-traditional or ontological relationship a slight distinction can be made depending on the domain at hand.
In the food safety ontology, the subject experts used the option to define their own relationships, e.g.:
compose: smaller parts making up a piece, element or sub-body of a bigger part
follow: to come before in order, time or position
The expected benefits of taking this approach were to provide support to the end-user in refining their search, and to facilitate navigation within the domain. Furthermore, an ontology editor was used to store the ontology in a database or in Resource Description Framework Specifications (RDFS).[15] The ontology can be accessed through a web interface; this allows the user to browse the ontology hierarchically or to search for a specific term and its relationships.
· Methodology
Ontologies emerged several years ago and the involvement of information specialists and the agricultural community is considerable. Over the last few years, great efforts were made to develop tools to facilitate ontology creation and a good theoretical basis. Despite all these efforts there is no well-defined framework available to date. Creating an ontology still involves a trial and error methodology, which may ultimately be the best approach. Additionally, its success also depends to a large extent on available financial and information resources as well as on the chosen domain. Former attempts to build an ontology can only rarely be reused because they are frequently too general and abstract, and therefore not very useful for the creation of concrete ontologies. This lack of methodology and expertise means that the process is invariably slow and can become complicated.
· Merging of thesauri and ontologies
The relationships in traditional thesauri (Broader Term, Narrower Term, and Related Term) are defined in the same way and do not cause any particular problems. Risks were commonly encountered when merging thesauri relate to the location of the descriptors in the hierarchy. The FAO and the National Agricultural Library (NAL) thesauri both use the descriptor "animal products".[16] However, the hierarchy related to this term differs in various points which is illustrated in the following table: Table 2: Comparing the same descriptor in two different thesauri
A particular characteristic of ontologies is the option to create one's own relationships in order to add more information. This can be an advantage, especially in clarifying the link between two terms. However, the freedom of not having to use standardized relationships can become problematic when ontologies are then merged. In these situations, subject specialists have to deal with the problem of merging the relationships correctly.
· Need for subject experts
The availability of subject specialists in ontology creation is crucial. Subject specialists have to be experts in the domain in which the particular ontology will be developed because only they are able to decide on the correct terms and relationships among those terms.
Although every subject specialist is an expert in his/her area, the understanding of a term might differ. The experts, therefore, have to pay attention that each of them uses the term under discussion in the same way. Where experts from related, yet slightly different, areas are involved it might even happen that they use the same term but with a completely different meaning depending on the context. Deciding - and agreeing - on relevant terms, their location within the hierarchy and the establishment of the relationships is a very time-consuming task for the experts and should not be underestimated.
