Hierarchical Methods and Apparatus for Extracting User Intent from Spoken Utterances

Abstract

Improved techniques are disclosed for permitting a user to employ more human-based grammar (i.e., free form or conversational input) while addressing a target system via a voice system. For example, a technique for determining intent associated with a spoken utterance of a user comprises the following steps/operations. Decoded speech uttered by the user is obtained. An intent is then extracted from the decoded speech uttered by the user. The intent is extracted in an iterative manner such that a first class is determined after a first iteration and a sub-class of the first class is determined after a second iteration. The first class and the sub-class of the first class are hierarchically indicative of the intent of the user, e.g., a target and data that may be associated with the target. The multi-stage intent extraction approach may have more than two iterations. By way of example only, the user intent extracting step may further determine a sub-class of the sub-class of the first class after a third iteration, such that the first class, the sub-class of the first class, and the sub-class of the sub-class of the first class are hierarchically indicative of the intent of the user.

1 . A method for determining intent associated with a spoken utterance of a user, comprising the steps of:
 obtaining decoded speech uttered by the user; and   extracting an intent from the decoded speech uttered by the user, wherein the intent is extracted in an iterative manner such that a first class is determined after a first iteration and at least one sub-class of the first class is determined after at least a second iteration, the first class and the at least one sub-class of the first class being hierarchically indicative of the intent of the user.   
     
     
         2 . The method of  claim 1 , wherein the user intent extracting step further comprises determining at least one sub-class of the sub-class of the first class after at least a third iteration, the first class, the sub-class of the first class, and the at least one sub-class of the sub-class of the first class being hierarchically indicative of the intent of the user. 
     
     
         3 . The method of  claim 2 , wherein the first class represents a target associated with the user intent, the sub-class of the first class represents an action associated with the target, and the sub-class of the sub-class of the first class represents data associated with the action. 
     
     
         4 . The method of  claim 1 , wherein the intent extracting step further comprises selecting a predetermined number of top scoring labels at each iteration. 
     
     
         5 . The method of  claim 1 , wherein the intent extracting step further comprises utilizing confidence scores at each iteration. 
     
     
         6 . The method of  claim 1 , wherein the intent extracting step further comprises utilizing a rejection criterion at each iteration. 
     
     
         7 . The method of  claim 1 , wherein one or more commands are provided to a target system based on the determination of the first class and the at least one sub-class of the first class. 
     
     
         8 . Apparatus for determining intent associated with a spoken utterance of a user, comprising:
 a memory; and   at least one processor coupled to the memory and operative to: (i) obtain decoded speech uttered by the user; and (ii) extract an intent from the decoded speech uttered by the user, wherein the intent is extracted in an iterative manner such that a first class is determined after a first iteration and at least one sub-class of the first class is determined after at least a second iteration, the first class and the at least one sub-class of the first class being hierarchically indicative of the intent of the user.   
     
     
         9 . The apparatus of  claim 8 , wherein the user intent extracting operation further comprises determining at least one sub-class of the sub-class of the first class after at least a third iteration, the first class, the sub-class of the first class, and the at least one sub-class of the sub-class of the first class being hierarchically indicative of the intent of the user. 
     
     
         10 . The apparatus of  claim 9 , wherein the first class represents a target associated with the user intent, the sub-class of the first class represents an action associated with the target, and the sub-class of the sub-class of the first class represents data associated with the action. 
     
     
         11 . The apparatus of  claim 8 , wherein the intent extracting operation further comprises selecting a predetermined number of top scoring labels at each iteration. 
     
     
         12 . The apparatus of  claim 8 , wherein the intent extracting operation further comprises utilizing confidence scores at each iteration. 
     
     
         13 . The apparatus of  claim 8 , wherein the intent extracting operation further comprises utilizing a rejection criterion at each iteration. 
     
     
         14 . The apparatus of  claim 8 , wherein one or more commands are provided to a target system based on the determination of the first class and the at least one sub-class of the first class. 
     
     
         15 . An article of manufacture for use in determining intent associated with a spoken utterance of a user, comprising a machine readable medium containing one or more programs which when executed implement the steps of:
 obtaining decoded speech uttered by the user; and   extracting an intent from the decoded speech uttered by the user, wherein the intent is extracted in an iterative manner such that a first class is determined after a first iteration and at least one sub-class of the first class is determined after at least a second iteration, the first class and the at least one sub-class of the first class being hierarchically indicative of the intent of the user.