Research on Common knowledge acquisition based on vision and language
Kei Matsubayashi ($B>>NS(B $B7D(B)
In a real world, man's knowledge is not in common.
Similarly, knowledge is not in common even on computer.
Knowledge is not completely shared between agents on computer when the
world in which they live is large, as well as between computer and
person who uses it.
But man can acquire common knowledge through multiple source of
This method makes realize common knowledge on computer.
Therefore, the purpose of this paper is to acquire common knowledge
between speaker and hearer making use of both visual information and
linguistical information as the example of some informations.
Common knowledge which this paper treats is the correspondence between
the word speaker used in order to represent the same object shared in
their scene and the one hearer did.
Hearer can acquire it comparing her/his own visual information with
But the viewpoint from which they view their scene is not always the same.
It is necessary to incorporate the viewpoint into the acquisition.
In this paper, we use situation semantics and propose the common
knowledge acquisition by adding
the comparison with visual information to formal treatment in
the semantics of perspectival expression -- the expression made
from a point of view.
Also, we implement the experimental system simplifying that treatment.
Then, we show it possible to acquire the common knowledge with
Abstract in Japanese:
How to get this paper?
To obtain this paper, please contact Kei Matsubayashi(firstname.lastname@example.org).
Thank you very much for your attention.