Inquiry Chain Cascade
An exploration episode in which one AI question leads to a chain of follow-up questions that draws the user progressively deeper into a topic, analogous to a 'Wikipedia hole' but driven by AI dialogue. The referent is the extended, self-propagating inquiry episode.
Operational Definition
EVENT. One occurrence is recorded when a single seed question develops into an uninterrupted chain of topically linked follow-up turns exceeding a set depth threshold (e.g. five or more consecutive on-topic follow-ups) within one session. The observable marker is a contiguous follow-up chain of qualifying depth descending from one initial query.
Measurement Schema
Proposed measurement protocol (not yet empirically validated): Session follow-up-chain analysis: dialogue turns are segmented into topic-linked chains; an episode would be coded when chain depth crosses the pre-set threshold. Could be reported as chains-per-session, mean and maximum chain depth, and chain dwell time; topic-linkage decisions corroborated by inter-turn semantic similarity and double-coded for reliability via Cohen's kappa.