Information Technology Reference
In-Depth Information
knowledge can be organized as a hierarchical structure. But which hierarchical
structure can reflect prior knowledge more intuitively?
Quillian [13,14,15] pointed out that prior knowledge is stored in the form
of semantic networks in human brain. A semantic network or knowledge struc-
ture is created with three primitives: concepts, relations, and instances. Yao [16]
pointed out human thought and knowledge is normally organized as hierarchical
structures, where concepts are ordered by their different levels of specificity or
granularity. A plausible reason for such organizations is that they reflect truth-
fully the hierarchical and nested structures abundant in natural and artificial
systems. Human perception and understanding of the real world depends, to a
large extent, on such nested and hierarchical structures.
Of course, we don't know the genuine organization of prior knowledge in
human brain until now. But from the above we know that prior knowledge
should be organized as a nested and hierarchical structures, which is helpful to
human problem solving.
2.3
Organization of Prior Knowledge
Maybe there have a lot of knowledge stored in human brain, but only a little part
of which is relevant to the solved problem in problem solving. So we should only
extract the relevant part of prior knowledge and reorganize them as a nested
and hierarchical structure in problem solving. Hierarchical structures not only
make a complex problem more easily understandable, but also lead to ecient
solutions.
In this subsection, we will provide a nested and hierarchical organization of
prior knowledge relevant to the solved problem, which is suitable for problem
solving particularly. We will reorganize relevant prior knowledge from one point
of view with multiple levels of granularity as a concept hierarchy, and reorganize
relevant prior knowledge from multilevel and multiview as a multidimensional
data model.
The formal use of concept hierarchies as the most important background
knowledge in data mining was introduced by Han, Cai and Cercone [18]. And
a multidimensional data model can be regarded as a combination of multiple
concept hierarchies. So it is rational to represents prior knowledge as concept
hierarchies or multidimensional data model.
Concept hierarchy.
Concepts are the basic unit of human thoughts and play
a central role in our understanding of the world. Human usually has a rich
clustering of concepts for knowledge from his familiar field, in which each concept
is related to many other concepts, and the relationships between concepts are
clearly understood. Concepts are arranged hierarchically using umbrella concepts
to more tightly relate them. The concept hierarchy is such an example.
In what follows, an organization of prior knowledge from one point of view
with multiple levels of granularity is proposed based on concept hierarchy. In
order to satisfy the need of nested and hierarchical structure, we will organize
relevant prior knowledge as a concept hierarchy in the form of tree in this paper.
