Developing an Intelligent Computational Model for Human Episodic Learning

Abstract

Human Episodic Learning (HEL) is a cognitive process which is a change in behavior that occurs as a result of an event. It is so-named because events are recorded into human episodic memory, which is critical for providing a memory of previous autobiographical events that can be explicitly stated. The principal mechanism for HEL as a mental process is an analogy. In which humans automatically learn new experienced knowledge (events) from different social dealings -which involve common social situations- by reliably selecting and reusing relevant prior socio-cultural knowledge (events) fragments, in the correct context, stored in the human episodic memory. Although there are many rich works done for computationally representing and capturing episodic knowledge. They suffer from some shortcomings due to neglect the context of the current social situation, having general knowledge about situations rather than rich descriptions of events flow, and difficulty in reasoning with used knowledge representations.So, in this thesis, a computational model for the human episodic learning process based on sociocultural situations is proposed in an attempt to overcome these shortcomings. The proposed model mimics how humans automatically learn by analogy robust and believable cognitive script-like knowledge from social dealings.