Information Technology Reference
In-Depth Information
piece of story (initially empty) and present as output a suggestion on how to continue
the story. Their main function is to effect corrections so the story plot can develop
correctly. If some inconsistency is detected in the input, the manager intervenes in
order to ensure that the original goal can be reached, satisfying all the necessary
constraints.
To model a chosen environment to which the plots to be composed should belong,
it is necessary to specify at least: a) what can exist at some state of the underlying
world; b) how states can be changed; and c) the factors driving the characters to act.
Accordingly, LOGTELL defines three schemas: static, dynamic and behavioural,
representing, respectively: a)
Storyworld:
the world description takes the form of a
set of facts (state), introducing the characters and their initial situation (relationships
and attributes); b)
World State Changes:
narratives are composed of events and an
event is a transition from a valid world state to another, which should also be valid;
and c)
Character's behaviour:
to model the reasons for each character's actions,
LOGTELL uses goal-inference rules specifying (in a logic formalism) the motivations
of these agents when certain situations occur.
State transitions consist of a limited repertoire of pre-defined domain-oriented
operations. A similar notion has been proposed by Vladimir Propp [16]. LOGTELL
equates the notion of event with the state-change brought about by the execution of an
operation by some agent. The dynamic schema is thus composed of a repertoire of
pre-defined operations (typical of the chosen genre) in which characters can take part.
The generation of a plot starts by inferring goals for the characters from the initial
configuration. A planner inserts events in the plot in order to allow the characters to
try to fulfill their goals. The result is then presented to the user. If the user accepts it,
the process continues by inferring new goals. The process alternates goal-inference
and plan generation until the user decides to stop or no new goal is inferred.
To further explore the possibilities of generating varied coherent stories in the
given “fairy-tale genre”, we enabled mechanisms also for plan-recognition.
Experiments with the system demonstrate that combining plan-recognition/generation
and user participation constitutes a promising strategy towards the production of plots
which are both diverse and coherent, as it helps guide the user [8]. However, that does
not guarantee dramatic power. Sound methods to combine events must be considered
to accomplish this goal; which led to the necessity to further extend the utilized
model. We do not claim that the present effort is sufficient to create a fully immersive
experience. We endeavour, as a more limited objective, to explore the possibilities of
generating a large variety of stories by applying this strategy over a given genre.
4
Event Relations in Plot Composition
Many narratologists agree that the most important relation between events is the
causal one. Nevertheless, this is not the only relation taken into consideration when
human beings try to create story patterns in their minds [14]. The impact on the
audience very much depends on how events are combined. Furtado [6] suggests that
at least four concerns are involved in plot composition: a) the plot must be formed by
