Information Technology Reference
In-Depth Information
OntoWiki [15], OpenRecord [16] and SweetWiki [17] are examples of wikis
where content is replaced by metadata.
2.3 The Adopted Approach
The adopted approach described in this chapter for semantic retrieval is based on
two main resources: an ontology and an algorithm derived from the classic vector
model [18].
An ontology is a means to formally model a domain of interest [19]. It consists
of a hierarchy of concepts, named classes, and associations between concepts.
Both classes and associations can be instantiated.
There are not so many useful ontologies as the research community would like
[20], because of unresolved technical limitations or the inexistence of sound ra-
tionales for why individuals refrain from building them; maybe due to the fact that
an ontology should result from the common understanding of a community on the
domain of interest.
Traditionally, an ontology follows a social concept of being the result of an
agreement on the understanding of a domain by some community.
Although, it is originally intended to make an explicit commitment to shared
meaning among an interested community, individuals can use ontologies to de-
scribe their own data [4]. For the considered approach, the ontology is not a
mechanism for communication and understanding of different agents but an adap-
tive information mechanism for a specific enterprise. Even though, a common on-
tology could be used as fine as a personal one.
The intended ontology represents the specific perspective of an organization, on
the domain where the wiki subject belongs regarding its interests. The kind of
adaptive behavior that is intended is also named personalization, when a determine
user is considered during her interactivity with the system.
Most systems with personalized behavior are based on some type of user pro-
file, such as an ontology in the OBIWAN project [6] and in the MESH project
[21]. The intended ontology is also used for adaptive behavior, without a specific
user, and should be build by the organization carefully, although explicit profile
creation is not recommended to avoid a burden on an individual user.
Ontologies enable the formalization of preferences in a common underlying, in-
teroperable representation, where interests can be matched to content meaning [21].
Differently from semantic wikis, the adopted approach is intended to extract in-
formation from plain wikis in a semantic way based on an ontology. The tool de-
scribed in this chapter performs retrievals from already edited wikis. There are no
changes in the way wikis are edited. There is no need to edit annotations and in-
clude them in text passages.
For ontology representation, it was chosen the Ontology Web Language (OWL)
[22], based on the Resource Description Framework (RDF), and it was used Pro-
tégé [23] as an ontology editor.
In the implemented tool, considering a wiki content, it is possible to process
slang, along with not formal and well written text. Comparable efforts do not sup-
port such functionality, as far as it is known.
