Contextual language generation by leveraging language understanding

摘要: Technology is provided for improving digital assistant performance by generating and presenting suggestions to users completing a task or session. To generate the suggestions, machine learned language prediction model trained with features extracted from multiple sources, such as log data session context. When input received user, used determine most likely suggestion present user lead successful completion. In suggestion, intermediate data, domain, intent, and/or slot, generated suggestion. From surface form of that can be presented user. The resulting related context may further continue training model.