Scaling statistical language understanding systems across domains and intents

摘要: A scalable statistical language understanding (SLU) system uses a fixed number of models that scale across domains and intents (i.e. single vs. multiple per utterance). For each domain added to the SLU system, existing is updated reflect newly domain. Information already included in corresponding training data may be re-used. The include detector model, an intent action object model slot/entity tagging model. identifies different identified within utterance. All/portion detected are used determine associated actions. determined action, one or more objects identified. Slot/entity performed using domains, actions, detector.