Authors: Yujie Zhang, Kennichi Kikuchi, Asuka Terai, Luning Ruan and Masanori Nakagawa
Abstract: We constructed two computational models of inductive reasoning based on a statistical analysis of Japanese corpora. One model applies Euclidean distance in its similarity function, whereas the other model uses Kullback–Leibler (KL) distance. Examining the similarity functions is crucial in a computational model of inductive reasoning. We used a correlation between psychological experiment and model simulation to compare these two models. KL distance is frequently used to measure the difference between two probability distributions. However, the results of this study indicate that a model using Euclidean distance is superior to a model using KL distance. More studies are required to determine the reason for this result.
Keywords:inductive reasoning; computational model; Euclidean distance; Kullback–Leibler distance