System and method of semi-supervised learning for spoken language understanding using semantic role labeling

摘要: A system and method are disclosed for providing semi-supervised learning a spoken language understanding module using semantic role labeling. The embodiment relates to of generating module. Steps in the comprise selecting at least one predicate/argument pair as an intent from set most frequent pairs domain, labeling training data mapping rules associated with selected pair, call-type classification model labeled data, re-labeling iteratively several above steps until labels converge.