One of the most important aspects of human reasoning is our ability to robustly adapt to new situations, tasks, and domains. Current AI systems exhibit brittleness when faced with new situations and domains. This work explores how structure mapping models of analogical processing allow for the robust reuse of domain knowledge. This work focuses on two analogical methods to reuse existing knowledge in novel situations and domains. The first method, analogical model formulation, applies analogy to the task of model formulation. Model formulation is the process of moving from a scenario or system description to a formal vocabulary of abstractions and causal models that can be used effectively for problem-solving. Analogical model formulation uses prior examples to determine which abstractions, assumptions, quantities, equations, and causal models are applicable in new situations within the same domain. By employing examples, the range of an analogical model formulation system is extendable by adding additional example-specific models. The robustness of this method for reasoning and learning is evaluated in a series of experiments in two domains, everyday physical reasoning with sketches and textbook physics problem-solving. The second method, domain transfer via analogy, is a task-level model of cross-domain analogical learning. DTA helps overcome brittleness by allowing abstract domain

an episodic memory for companion dialogue

Artificial Companions are computer systems that interact and collaborate with the user over a longer period of time, employing human-like communication. They have memory and are able to learn from the interaction with their users. Due to their long-term relationships with users and their ability to communicate in a human-like fashion they have social effects on the user. We aim at improving the dialogue capabilities of companions by endowing them with episodic memory (EM) and integrating it with action selection and dialogue control. Our research builds upon the platform and companions developed in the RASCALLI project [1][2], but can be easily transferred to other applications. Goals The goal of equipping companions with EM is to achieve more coherent, believable behaviour based on experiences: support action selection based on similarity to past experiences explain actions of companions (mistakes / offline tasks) provide more interesting dialogue by being able to make statements about: users ' current and past preferences, and their relation to the current situation patterns, similarities and dissimilarities between current and past interactions employ user preferences stored in the EM to: improve answer retrieval from knowledge bases proactively present interesting information to the user Method We propose an episodic memory component integrated with action selection and dialogue control. Episodic memory is part of long-term memory. It stores personal experiences in association with their time, mindset, feelings, and place [3]. Models of EM have been used successfully in robotics [4], artificial life agents [5], interactive storytelling [6], and non-player characters [7]. While our use of episodic memory is tightly integrated with the agent's dialogue control, the EM is still taskindependent and can be used by other components.