QUEST: A cognitive model of question answering

Publisher Website
Google Scholar
Abstract
QUEST is a cognitive model of question answering that accounts for the answers adults give to questions such as “Why?” “How?” “When?” “Where?” “What are the consequences of X?” “Is X true or false?” QUEST specifies the information sources that are tapped for answers to a question. Each information source is a highly structured data base, with statement nodes organized by a network of directed relational arcs. The types of knowledge structures include causal networks, goal hierarchies, taxonomic hierarchies, and spatial networks. QUEST'S convergence mechanisms identify the subset of statement nodes within an information source that furnish good answers to a question. The convergence mechanisms include (1) arc‐search procedure for each question category, (2) identification of intersecting nodes from different information sources, (3) constraint propagation, and (4) pragmatic considerations such as the goals and common ground of speech participants. QUEST accounts for question‐answering data in several psychological tasks in both laboratory and naturalistic settings. Parts of QUEST have been simulated by computer.