BEZERRA, H. C. F.; BEZERRA, Haroldo Cesar Frota.
Résumé:
The critical phase of the process of building expert systems is the
responsible for generating the knowledge base because she is expensive
difficult and greatly influences the final quality of these systems.
Induction from examples generates knowledge bases in
decision trees or rules through algorithms called
inductive algorithms.
Inductive algorithms can be non-incremental or incremental algorithms.
Non-incremental algorithms are input a set of
examples and output as a knowledge base.
Incremental algorithms receive as input a database of
knowledge and a set of new examples.
updated knowledge base.
In this paper we present a comparative performance analysis
between these two families of decision tree generating algorithms.
We show through theoretical and experimental comparative analysis that,
contrary to what the authors of the incremental algorithms said
these algorithms, only in rare domain and domain situations.
number of examples these algorithms were more efficient
non-incremental, thus not meeting the objectives
proposed.
