ABSTRACT
When individuals answer questions, they tap multiple sources of information that embody world knowledge. We assume that each information source is a structured database containing content nodes that are connected by relational arcs. Structured databases are incorporated in various types of information systems, such as intelligent tutoring systems, expert systems, and decision support systems. We also assume that there are a variety of question-answering strategies that produce good answers by accessing particular information sources and systematically selecting nodes within each information source. A psychological theory of question answering would (at minimum) document the question-answering strategies that are associated with different categories of questions.
