Learning cross-domain social knowledge from cognitive scripts

Creating artificial intelligence (AI) agents with computational models based on human cognitive abilities is an ongoing research area. This paper proposes a new evolutionary cognitive model for cross-domain learning, which aims to improve the cognitive learning process by extracting new experienced knowledge from pre-existing socio-cultural cognitive scripts. This knowledge is necessary for the Al agents to develop learning for the current faced social situation (Target). The model depends basically on two phases; the retrieval phase and the learning phase. In the retrieval phase, Pharaoh algorithm is utilized to retrieve the most relevant cognitive script (Base) to the target script considering the context. Whereas, the learning phase employs evolutional processes to enrich the target script. Finally, the enriched script replaces the target script in the evolved script-base in order to be used in the learning and retrieval phases again.