The present invention relates to a method for updating language understanding classifier models which includes receiving via one or more microphones from a computing device, a digital voice input from a computing device user. Natural language processing using digital voice input is used to determine a user's voice request. In determining that the user's voice request does not match at least one of several predefined voice commands in a schema definition of a digital personal assistant, an end-user marking tool gui is used to receive a user selection. of at least one of the following: at least one intention from several available intentions and / or at least one attribute for at least one intention. A tagged data set is generated by matching the user's voice request and user selection, and is used to update a language understanding classifier.
展开▼