Semantic Web Service Usage Tasks in WSMO - Enabling Semantic Web Services: The Web Service Modeling Ontology - page 115

Keyword-Based Discovery
Keyword-based discovery is a basic ingredient in a complete framework for
Semantic Web service discovery. By performing a keyword-based search, the
huge amount of available services can be filtered or ranked rather quickly.
In a typical keyword-based scenario, a keyword-based query engine is used
to discover services. A query, which is basically a set of keywords, is provided
as input to a query engine. The query engine matches the keywords in the
user's input against the keywords used to describe the service. A query with
the same meaning can be formulated by using a dictionary of synonyms, such
as WordNet [37]. The semantics of the query remains the same but, because of
the different keywords used (synonyms of previous ones), more services that
possibly fulfill the user's request are found. Moreover, by using dictionaries
such as WordNet, as well as natural-language processing techniques, an in-
crease in the semantic relevance of the search results (with regard to the search
request) can in principle be achieved [115]. Nonetheless, such techniques are
inherently restricted by the ambiguities of natural language and the lack of
semantic understanding of natural-language descriptions by algorithmic sys-
tems.
The services descriptions at least consist of a dedicated set of keywords
for categorizing and indexing. Furthermore, they can include richer, semantic-
based elements.
Discovery Based on Simple Semantic Descriptions
Although keyword-based search is a widely used technique for information
retrieval, it does not use explicit, well-defined semantics. The keywords used
to retrieve relevant information do not have an explicit formalization and,
therefore, do not allow inferencing to improve the search results.
For these reasons, as a second approach, we consider the use of controlled
vocabularies with explicit, formal semantics. Ontologies provide excellent and
prominent conceptual means for this purpose. They provide an explicit and
shared terminology, and explicate interdependencies between concepts, and
thus are well suited to the description of Web services and goals. Moreover,
ontologies can be formalized in logics, which enables the use of inference ser-
vices for exploiting knowledge about the problem domain during discovery.
Web services at this intermediate level of abstraction, are understood as
sets of (unstructured) elements, i.e. as concepts from an ontological perspec-
tive. A capability of an Web service at this level of abstraction is a description
which does not depend on dynamic factors. A capability describes only what
an advertised Web service can potentially deliver and not under which circum-
stances (meaning preconditions) particular services can actually be provided.