No label defined (Q4382283)

The book presents ideas on how to help people learn, or, more specifically, how to learn interactively. The main idea is to simulate real-life systems (e.g. a coffee maker) on a computer. A learner explores the behavior of the system by playing with it or rather with its computer model. Such learning is safe and inexpensive. The learner needs some guidance during the process of learning, otherwise learning may become inefficient: the learner will play with the model (a qualitative simulator) not paying any attention to discovering the underlying principles. The problem is known as the ``science museum problem''. The book presents the STAR methodology for such guidance. Representations for modeling the knowledge necessary to predict the behavior of the system together with algorithms to generate these models automatically from the output of a model should be given. As a result, a subject matter model is created to predict behavior of the model, useful for learning. Additionally, techniques for automatic adding hierarchical structure are also created. Without hierarchization the model would be too complicated. The book is written in a causal, informal way. A plethora of examples are used for illustrating all ideas, including articulate models, inference graphs, aggregation techniques, diagnosis of learner's behavior. In summary, the book might be useful for designers of simulators of complex systems.