Information Technology Reference
In-Depth Information
Generally the heuristics in this method serves by giving evaluations on
approximate solutions and is controlled by rule-based production systems.
4. Derivational replay
The above three kinds of revise methods are based on adaptations on the
solutions to the base cases. Derivational replay cares how the solutions driven,
and use the method of deriving an old solution or solution piece to derive a
solution in the new situation. Compared with the case-based substitution
methods mentioned above, derivational replay is a kind of case-based revise
methods.
5.9 Case Retainion
The process of case retainion is similar to that of case retrieval. The word
“remember” has two kinds of meanings: storage and retrieval. Incorporating
the problem and solution to the case base as a new case involves the
index-mechanism to assign indices to the cases to facilitate its future retrieval,
and the insertion algorithms to store the new case in the appropriate place in
the case base. Generally speaking, the search performed in case retainion is
the same as that in case retrieval. The insertion algorithms aim to find a place
for the new case, while retrieval algorithms are to return the most similar
cases. Arrangement on cases begins when potential useful cases are returned,
and the systems need to reorganize the case base after the insertion of new
cases.
If the new problem has been solved successfully, the problem and solution
should be stored as a new case for future use. Case retainion is the process of
learning and knowledge acquisition, which involves deciding what
information to retain and in what form to retain it; how to index the case for
future retrieval; how to integrate the new case into the case base, and how to
organize and manage the case base.
Deciding what information to retain includes the following considerations:
Feature descriptions involved in the problem; the solution to the problem; and
explanations on the success.
Additional indices should be assigned to the newly-added cases to
facilitate their retrieval. Indices should be constructed in such a way that the
cases only be retrieved in the related cases. For this reason, adjustments to the
index contents, even the structure of case base should be done, such as
changes on the intensity of the indices or the weights of feature concerned.
