Information Technology Reference
In-Depth Information
Non-propositional fluents were rst introduced in [24]. The dialect A C al-
lows to represent and reason about the eects of concurrently performed
actions [4]. Non-deterministic actions are supported in A N , and the lan-
guage A NCC extends the latter and combines it with A C [12]. In AR ,
too, non-deterministic actions can be formalized and, moreover, it includes
a basic solution to the problem of indirect eects of actions [57, 40]. The
extension L 0 supports incomplete specication of the action sequence that
actually occurs in a scenario [6]. Information gathering actions can be repre-
sented in the language A k [70]. The language E of [55] is based on a linear
time structure and allows to represent narratives of events. A good general
introduction to the Action Description Language and several dialects is [35].
The second major class of action theories has its roots in the language
and semantics developed in [89]. A distinctive feature of this framework,
which is commonly called \Features-and-Fluents," is that it allows to rea-
son about the duration of actions. A thorough and ne-grained hierarchy of
sub-classes with restricted expressiveness allows precise assessment results
for dierent action formalisms. Major evaluation results are reported in [90].
Several extensions to the original framework have been developed, e.g., con-
currency of actions [122], actions with indirect eects [91, 47, 93], and goal-
orientedness [95]. The reference article [94] oers a general introduction to
this line of research.
A rst formal comparison between Action Description Language A and
\Features-and-Fluents" was established in [107], where both theories are
proved equivalent for a particular class of domains. Of course many more for-
malisms exist which arguably deserve being called action theories. The Action
Description Language and the \Features-and-Fluents" framework, however,
are the only ones that found broader dissemination in two respects: They
have been employed for the validation of a variety of action formalisms, and
they have been used as a uniform basis for dealing with a variety of onto-
logical aspects. Insofar as these and other more specialized action theories
address the problems we are concerned with in the following chapters, they
will be described and discussed in detail in due course.
Search WWH ::




Custom Search