Information Technology Reference
In-Depth Information
production rules, frames, and object-oriented formalisms. In some cases, this
makes knowledge processing can be done locally. In the memory network, the
semantic relations among nodes guarantee the easy retrieval of knowledge
relevant to a certain SMU.
It easy to see that the memory network is quite complicated, yet it can
really reflected the intricate inner relations among all kinds of knowledge.
The complexity of the network makes the construction of the memory
network and its learning process very complex too. For people, the memory
network is the result of long-term accumulation of studying and thinking. In
this process, new nodes and knowledge are increasing constantly; meanwhile,
the knowledge about a certain node not used for a long time is forgotten. It
means that the construction process of the memory network is a learning
process of knowledge in fact.
The memory network can account for the forgetfulness of knowledge in
some sense. Some concrete contents of the nearly-unused knowledge are
generally totally forgotten unconsciously, yet some roughly impressions about
this kind of knowledge can be left in memory. It means that the memory
network is a long-time memory, while the memory of slots is short-term and
will die away gradually or vanish. We can describe the above phenomenon
with the notation of memory intensity. In general, the memory intensity is a
function of time and recall. As time elapses, the memory intensity will be
weakened, and after a recall, the memory intensity of the knowledge increases
to some extent.
The memory network has some resemblance to the semantic network. It is
a kind of model developed on the basis of the semantic network; and they
both code information in networks: nodes to express information, arcs
between nodes to express the semantic relations. Yet there are many
differences between them. The most primary difference is in the expression of
information. The information expression power of the semantic network are
confined to the network itself, namely knowledge can only be represented by
nodes and the connections among nodes. But the expression power of the
memory network is far beyond the semantic network:
(1) Store and use the theory and concrete cases represented in other
representation methods;
(2) Store special knowledge by adding restraints to the nodes;
(3) Organize similar knowledge by intension nodes.
