Domain Specific Speech Acts for Spoken Language Translation

摘要: We describe a coding scheme for machine translation of spoken taskoriented dialogue. The covers two levels speaker intention − domain independent speech acts and dependent actions. Our database contains over 14,000 tagged sentences in English, Italian, German. argue that actions, not acts, are the relevant discourse unit improving quality. also show that, although actions specific, approach scales up to large domains without an explosion can be coded with high inter-coder reliability across research sites. Furthermore, number is on order ten times sparseness problem training classifiers identifying action. our work developing accuracy act action classifiers, which core source language analysis module NESPOLE system.