Neural symbolic architecture for normative agents

In this paper we propose a neural-symbolic architecture to representand reason with norms in multi-agent systems. On the one hand,the architecture contains a symbolic knowledge base to representnorms and on the other hand it contains a neural network to rea-son with norms. The interaction between the symbolic knowledgeand the neural network is used to learn norms. We describe howto handle normative reasoning issues like contrary to duties, dilem-mas and exceptions by using a priority-based ordering between thenorms in a neural-symbolic architecture.