Information Technology Reference
In-Depth Information
The desired relationship can be described most adequately as the se-
mantic difference between functional descriptions. For two given functional
descriptions φ and ψ ,the -relation between them is defined as φ,ψ =
( φ
ψ ), i.e. contains those elements that are models for either
φ or ψ and not common to them. A -relation defines a symmetric relation
between φ and ψ ; when we concatenate φ,ψ with either φ or ψ ,weobtain
logical equality with the corresponding other formula.
∧¬
ψ )
(
¬
φ
Beneficial Usage Scenarios.
While the benefits obtainable from data- and process-level mediation are obvi-
ous, the benefits of functional-level mediation may be less obvious. Therefore,
let us consider the following situation in the VTA scenario: G 1 defines a re-
quest to buy a ticket from Innsbruck to Vienna, WS 1 provides a purchase
facility for train tickets in Austria, and WS 2 provides a facility for booking
flight tickets in Austria and for international flights.
The -relations between G 1 and each of the Web services provides addi-
tional knowledge about the circumstances under which each Web service can
be used for resolving G 1 :using WS 1 will result in the purchased ticket being a
train ticket; using WS 2 will result in the purchased ticket being a flight ticket.
This additional information can for example be utilized for selecting the Web
service to be used. Imagine that the client who defines G 1 has a preference for
traveling by plane because of tight business schedules. Here, we would choose
WS 2 for use as the constraint of using WS 1 contradicts the client's preference.
Other beneficial mechanisms that work with -relations are:
Goal Adjustment. This is concerned with adjusting goal definitions that are
not resolvable a priori. Within our scenario, imagine that a client defines
a goal G 3 of buying a flight ticket from Innsbruck to Vienna for less than
e
200. The flight booking Web service WS 2 seems to be usable, but all
purchasable flight tickets cost significantly more than
200. However,
train tickets for the desired itinerary are available for less than
e
200 from
WS 1 . With respect to the -relation between G 3 and the Web services,
we can propose possibilities for adjusting the goal. Either weaken it by
omitting the price constraint so that WS 2 becomes usable, or change the
desired ticket type to train ticket so that WS 1 becomes usable. Note that
in any case the objective of the client as specified in G 3 will be changed,
so that interaction with the client is required for confirmation.
e
Goal Ontologies. Presumably, the most effective usage scenario with re-
spect to gainable e ciency increase would be to compute the -relations
between already resolved goals and Web services, and then to utilize this
additional knowledge to avoid or reduce the need for complex reasoning
procedures for functional discovery. For instance, we can omit runs of dis-
covery engines for determining usable Web services for the above-defined
goal G 2 if we already know WS 1 and WS 2 to be usable for G 1 ,asweobtain
from the -relation between G 1 and G 2 the fact that they are structurally
Search WWH ::




Custom Search