Abstract
In this paper, we propose a moral judgment system using distributed representation and associative information. Moral judgment means right and wrong judgment. There are two features in the proposed system. The first one is the usage of distributed representation of sentences. It enables consideration of the meaning of the complicated sentences. Here, the distributed representations adjusted for moral judgment are used to increase the accuracy. The second one is the usage of associative information to consider the related knowledge. In the proposed system, a database of predicate argument structure is constructed by a large corpus (British National Corpus) and is used for association. In the experiments, we compared the proposed system with the judgments by human. We confirmed the effectiveness of the proposed system.
