In the last few years CAD systems have been evolving from simple drafting tools to much more complet
solid modeling environments. Nevertheless, experience
has shown that an effective use of such systems relies
on the characteristics of their user interface: the user
should have the possibility of describing in full detail a
particular scenario or giving the system only a raw description of it. This paper describes NALIG, a system
able to Lnderstand” and “reason about” high level descriptions of spatial scenes. The user interacts with
the system by using a simple natural language fragment but expressive enough to describe complez configurations of objects. NALlG replies by drawing on
the screen an image mirroring its own ‘understanding” of the scene described. The comprehension process involves various forms of common sense reasoning
cam‘ed out at iwo different levels of abstraction. This
has required the integration of diflerent AI-techniques
(e.g. natural language understanding, spatial reasoning, default reasoning).
http://www.aleeshainstitute.com/interior-designing-course.php
The ability to free the user from the task of providing a detailed and complete specification seems one of
the most compelling requirements for flexible and intelligent (CAD) systems. This paper describes NALIG,
a system able to “understand” and “reason about”
high level descriptions of spatial scenes. The user describes the intended scenario by issuing NALIG a sequence of locative expressions predicating the position of an object relative to another. Such a descrip
tion drives the construction and updating of a spatial
representation which plays the role of NALIG’s “mental image” of the scenario [l]. In reply NALIG draws
onto the screen a graphical image mirroring the inner spatial representation. In such a way the user can
P. Pecchiari
Mechanized Reasoning Group
DIST - University of Genova
Genova, ITALY 16145
immediately recognize misunderstandings and correct
NALIG.
The sentences accepted by the system obey the following syntax:
sb j[n] prep ob j[m] [ref-sys]
where prep is one of the following spatial prepositions: on, in, over, under, behind, in front of, to the
right of, to the left of, near, far from. sbj is the name
of the located object and obj the name of the reference object. The (optional) specification of a third
object (ref-sys) may occur when a projective preposition (e.g. to the left of, to the right of) is used [3]. The
name of an object may be either an individual name
or a class name. A class name (e.g. chair) provides
information only relative to the object prototype of
which the referenced object is an instance. Individual
names (e.g. chair3) are clearly less problematic since
they unambiguously refer to distinguished objects.
The key feature of the system reside in the interplay between symbolic and numerical reasoning components embodied by the Sybolic Module and by the
Positioning Module respectively. The former being
mainly concerned with the manipulation of relational
knowledge while the latter with the processing of metric information. The Symbolic Module extracts from
the sentence given by the user a set of concepiualizaiions. Conceptualizations are ground predicates intended to unambiguously encode the spatial relations
between the objects. For instance, given the sentence
chair on ffoor, NALIG engenders the conceptualization
EXOITACT(chair2, floor). EXONTACT represents
the particular sense the preposition on has in the context of the previous sentence: the relation of horizontal
support provided by the reference object (i.e. floor)
to the located object (i.e. chair2). Notice that, if we
consider the sentence picture on wall, the same spatial
preposition (i.e. on) is used with a different intended
meaning and, according to such a different interpretation, NALlG generates VXONTACT(picture1, wall).
Comments
Post a Comment