Genetic algorithms for analogical mapping
Bjørnar Tessem · 2002
The paper presents an application of genetic algorithms in analogical reasoning. Analogical reasoning means to utilize structural and semantic similarities between relational descriptions of two situations to establish new knowledge in one of the two situations. The author exploits the properties of genetic algorithms as a method to solve discrete optimization problems with many local optima. The author focuses on the mapping phase of analogical reasoning, and represents this problem as an objective function to be optimized by genetic algorithms. Examples and experiments indicate that the approach has expressibility and efficiency comparable to previous analogical mapping algorithms.