Graphics Reference
In-Depth Information
directional and therefore does not necessarily apply in the opposite
direction. For initialization purposes, if two agents do not know each
other, a neutral initial value of 0 is assumed.
The importance of the goal is especially relevant for the individual
goals . It is represented for each defined goal with a value between 0
(no relevance or non-existent) and 1 (maximum relevance) (rel Goal =
[0, 1]).
Currently, COMPLEX only defines the actions of individual
agents as events . These events are categorized depending on the
individual values of each agent with regard to their praiseworthiness
(praiseworthiness Action = [-1, 1]). Consequently, this variable defines
whether an agent feels “ashamed” for performing an action, or feels
“pride” in the other case. The ' praiseworthiness' variable may range
between “1 (low value) and +1 (high value). Internal events (e.g.
psychophysiological parameters of the user) are (not yet) taken into
account in the current implementation of COMPLEX.
How an event is assessed within COMPLEX also depends on its
consequences (consequence Goal, Group, Prospect = [-1, 1]). Consequence is
understood as the influence on an individual goal. The consequence
variable may range between “1 (hindering the goal) and +1 (facilitating
the goal). In addition, the consequences of an action are appraised
separately for different groups (Group = [Self, Other, Concerned]) and
may also occur in the future and not at the time an action is carried
out (Prospect = [true, false]). Handing in a paper, for example, may
not lead to a feeling of deep satisfaction, but initially only the hope
that it is accepted. Satisfaction comes later, once the paper has been
accepted.
Consequently, an event or an action may have numerous
consequences. The definition of these individual consequences is very
difficult to implement in COMPLEX because the variables are domain-
specific and must be defined for every application domain. Especially
problematic are the variables ' praiseworthiness' and ' consequence'
because they cannot be defined globally for a domain, but are based on
individual values (such as, for example, stable or variable personality
traits) and are therefore agent-specific.
The direct output of the appraisal process, which uses the OCC
model, are discrete emotions. In terms of time, however, these last
milliseconds to seconds are therefore not long enough to model the
emotional experience over longer periods of time. COMPLEX tries
to solve this problem with the help of additional variables that last
longer, for example, moods (medium-term) or personality (long-
term). Both the emotional state and the mood are internally mapped
Search WWH ::




Custom Search