Exploratory Model Building
This addresses the need for agents to reason beyond past experiences in creative tasks, though it appears incremental as it builds on existing probabilistic knowledge.
The paper tackles the problem of creative thinking by developing a formalism for exploring conceivable situation-models in probabilistic domains, aiming to construct consistent, minimal, and desirable descriptions.
Some instances of creative thinking require an agent to build and test hypothetical theories. Such a reasoner needs to explore the space of not only those situations that have occurred in the past, but also those that are rationally conceivable. In this paper we present a formalism for exploring the space of conceivable situation-models for those domains in which the knowledge is primarily probabilistic in nature. The formalism seeks to construct consistent, minimal, and desirable situation-descriptions by selecting suitable domain-attributes and dependency relationships from the available domain knowledge.