VIRTUAL ASSISTANT DEVELOPMENT SYSTEM

• US 20160062604A1
• Filed: 08/29/2014
• Published: 03/03/2016
• Est. Priority Date: 08/29/2014
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Copperman, Max, Daniel, Susan Dawnstarr, Kraljic, Tanya, Cabaco, Tiago G.
• Granted Patent: US 10,698,585 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 3/0482 Interaction with lists of selectable items, e.g. menus
• CPC Class Codes: G06F 3/0483 Interaction with page-structured environments, e.g. book metaphor
• CPC Class Codes: G06F 3/04842 Selection of displayed objects or displayed text elements G06F3/0482 takes precedence
• CPC Class Codes: G06F 3/04847 Interaction techniques to control parameter settings, e.g. interaction with sliders or dials
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 2015/221 Announcement of recognition results

Abstract

In accordance with aspects of the disclosure, a computing device may provide a user interface for developing an interactive natural-language response system, which may include a virtual assistant. A user may interact with a system using spoken, written (e.g., text), or other input methods. The user interface may allow a user to associate sentences with intents, tag words within the sentences with concepts, and construct a grammar using the associated intents and tagged concepts. The system may use the grammar for automatically predictively associating sentences with intents and words with concepts. The system may display in the foam of a chat transcript a single branch of a tree of a discussion between the virtual assistant and a user. The user interface may graphically display variable values to assist a user to test system responses under different simulated conditions.

Claims

1. A non-transitory computer-readable medium storing computer-readable instructions that, when executed by a processor, cause a device to provide a user interface comprising:

a first display area that displays a list of one or more intents, wherein each intent of the one or more intents is associated with a process that can be performed by a language-processing system;

a second display area that displays a list of request strings, wherein the list of request strings is automatically updated, responsive to a selection of an intent from the list of one or more intents in the first display area, to include one or more request strings associated with the selected intent; and

a third display area that displays one or more concept tags respectively corresponding to one or more concepts respectively associated with one or more associated words in a request string of the list of request strings in the second display area, and receives input associating one or more unassociated words in the request string with one or more concepts, wherein the one or more concepts are associated with the selected intent, wherein the request string of the list of request strings in the second display area is associated with the selected intent.

2. The non-transitory computer-readable medium of claim 1, wherein the user interface comprises:

a fourth display area that receives a request string, displays an intent associated with the request string, and displays one or more concept tags respectively associated with one or more concepts respectively associated with one or more words in the request string, wherein the one or more concepts are associated with the intent associated with the request string.

3. The non-transitory computer-readable medium of claim 2, wherein the fourth display area automatically updates the one or more concept tags responsive to receiving input indicating a user-added association between a different intent and the request string.

4. The non-transitory computer-readable medium of claim 1, wherein the third display area automatically displays a new concept in a list of concepts associated with the selected intent responsive to receiving input adding the new concept to the one or more concepts associated with the selected intent.

5. The non-transitory computer-readable medium of claim 1, wherein the third display area is displayed in response to a selection of the request string of the list of request strings in the second display area, and wherein the third display area is hidden in response to a deselection of the request string of the list of request strings in the second display area.

6. The non-transitory computer-readable medium of claim 1, storing computer-readable instructions that, when executed by the processor, cause the device to:

receive a new request string not currently in the list of request strings;

suggest an intent of the list of one or more intents to associate with the new request string;

suggest one or more concept tags respectively corresponding to one or more concepts to respectively associate with one or more words in the new request string, wherein the one or more concepts are associated with the suggested intent; and

display the new request string, the suggested intent, and the suggested one or more concept tags in the user interface.

7. The non-transitory computer-readable medium of claim 6, storing computer-readable instructions that, when executed by the processor, cause the device to:

receive validation of the suggested intent; and

associate the new request string with the suggested intent.

8. The non-transitory computer-readable medium of claim 6, storing computer-readable instructions that, when executed by the processor, cause the device to:

receive validation of the suggested one or more concept tags; and

associate the suggested one or more concept tags respectively corresponding to the one or more concepts with the one or more words in the new request string.

9. The non-transitory computer-readable medium of claim 6, storing computer-readable instructions that, when executed by the processor, cause the device to:

generate a grammar using the request strings, the one or more intents, and the one or more concepts,

wherein the suggesting the intent of the list of one or more intents to associate with the new request string is based on the grammar, and

wherein the suggesting the one or more concept tags respectively corresponding to the one or more concepts to respectively associate with the one or more words in the new request string is based on the grammar.

10. The non-transitory computer-readable medium of claim 1, storing computer-readable instructions that, when executed by the processor, cause the device to:

dissociate the request string of the list of request strings from the selected intent in response to one or more inputs received from a user of the user interface; and

associate the request string with a different intent than the selected intent in response to the one or more inputs received from the user.

11. A non-transitory computer-readable medium storing computer-readable instructions that, when executed by a processor, cause a device to:

receive a tree comprising a plurality of discussions between a virtual assistant and a user, each discussion of the plurality of discussions being represented by a discussion path of the tree, each discussion path comprising one or more nodes, each node of the one or more nodes associated with one of a plurality of potential user responses that the user may give the virtual assistant or one of a plurality of potential system responses that the virtual assistant may give the user;

generate a display of a transcript comprising a single discussion path representing one discussion of the plurality of discussions, the transcript comprising a series of interactions between the virtual assistant and the user, wherein each interaction of the series of interactions is associated with a respective node of the one or more nodes;

receive a selection of a node of the one or more nodes;

generate a display of the plurality of potential user responses respectively associated with a plurality of next nodes after the selected node;

receive a selection of a user response of the displayed plurality of potential user responses;

update the display of the transcript to show the selected user response; and

update the display of the transcript to show the one or more system responses of the plurality of potential system responses, wherein the one or more system responses is associated with a next node of the plurality of next nodes after the selected node.

12. The non-transitory computer-readable medium of claim 11, wherein the single discussion path is determined by traversing the tree by taking one of a first branch at each node, a most heavily traveled branch at each node, and an arbitrary selection of a branch at each node.

13. The non-transitory computer-readable medium of claim 11, wherein each node of the one or more nodes is associated with one or more of contextual information retrieved from another source, contextual information received during a configuration process, and contextual information previously determined and stored.

14. The non-transitory computer-readable medium of claim 11, storing computer-readable instructions that, when executed by the processor, cause the device to:

receive a starting string;

create a top node of a tree using the starting string;

receive a response string that is responsive to the starting string;

create a next node using the response string; and

create a connection between the top node and the next node, wherein the connection between the top node and the next node comprises a branch of the tree.

15. The non-transitory computer-readable medium of claim 14, wherein the tree comprises a plurality of nodes, and wherein the computer-readable instructions, when executed by the processor, cause the device to:

receive a selection of a node of the tree;

receive a response string associated with the selected node of the tree;

create a new node using the response string; and

create a connection between the selected node and the new node, wherein the connection between the selected node and the new node comprises a new branch of the tree.

16. The non-transitory computer-readable medium of claim 11, wherein the tree comprises a plurality of trees,

wherein each tree of the plurality of trees is each associated with a function of the virtual assistant,

wherein a top node of each tree is associated with a potential user response of a plurality of potential user responses.

17. The non-transitory computer-readable medium of claim 11, wherein a top node of the one or more nodes is associated with an intent, and the intent is associated with the one of the plurality of potential user responses or the one of the plurality of potential system responses associated with the top node,

wherein a grammar recognizes a request string as matching the intent by matching the request string to the one of the plurality of potential user responses or the one of the plurality of system responses associated with the top node.

18. The non-transitory computer-readable medium of claim 11, wherein the selected node is associated with a concept, and the plurality of next nodes after the selected node are respectively associated with a distinct value of the concept,

wherein the distinct value of the concept associated with a next node of the plurality of next nodes is recognized by a grammar as matching the potential user response associated with the one node.

19. A non-transitory computer-readable medium storing computer-readable instructions that, when executed by a processor, cause a device to:

receive, via a graphical interface, an input string;

receive an association of the input string with an intent of a virtual assistant;

create a top node of a tree;

associate the top node of the tree with the input string and the intent of the virtual assistant;

receive, via the graphical interface, a response string that is responsive to the input string;

create a next node after the top node of the tree;

associate the next node with the response string;

receive, via the graphical interface, a different response string that is responsive to the input string;

create a different next node after the top node of the tree; and

associate the different next node with the different response string,

wherein the graphical interface hides one of the response string and the different response string in response to displaying another of the response string and the different response string.

20. The non-transitory computer-readable medium of claim 19, wherein a grammar recognizes a request string as matching the intent by matching the request string to one of a plurality of response strings associated with the top node.

21. A non-transitory computer-readable medium storing computer-readable instructions that, when executed by a processor, cause a device to:

display a testing interface comprising a graphical representation of a plurality of variables associated with a particular function of an virtual assistant, the particular function being associated with an intent of the virtual assistant, and each variable of the plurality of variables corresponding to a concept associated with the intent;

receive input, wherein the device determines the particular function associated with the intent of the virtual assistant based on the input; and

display a visual representation of a hypothetical dialog between a user and the virtual assistant, wherein the hypothetical dialog comprises the input.

22. The non-transitory computer-readable medium of claim 21, storing computer-readable instructions that, when executed by the processor, cause the device to:

receive a different input, wherein at least one of the input and the different input comprises a value of at least one concept of the concepts associated with the intent;

set a variable corresponding to the at least one concept to the value, wherein the variable is one of the plurality of variables; and

update the graphical representation of the plurality of variables to include the value.

23. The non-transitory computer-readable medium of claim 22, wherein the visual representation of the hypothetical dialog between the user and the virtual assistant comprises a display of a transcript of a single potential path through a tree, the tree comprising a plurality of discussions between the virtual assistant and the user,

wherein the single potential path through the tree is determined based on the value of the variable.

24. The non-transitory computer-readable medium of claim 23, storing computer-readable instructions that, when executed by the processor, cause the device to:

hide a potential system response that the virtual assistant may give the user in response to determining that the potential system response is associated with the variable and the variable is set to the value, wherein the single potential path through the tree comprises the potential system response.

25. The non-transitory computer-readable medium of claim 21, storing computer-readable instructions that, when executed by the processor, cause the device to:

receive, from a user of the testing interface, a different input comprising a user response; and

determine, based on the different input comprising the user response, a plurality of questions and responses.

26. The non-transitory computer-readable medium of claim 21, storing computer-readable instructions that, when executed by the processor, cause the device to:

call a running of the virtual assistant;

provide the virtual assistant with a plurality of values of the plurality of variables;

provide the virtual assistant with a plurality of hypothetical user responses; and

receive, from the virtual assistant, one or more questions to ask the user in the hypothetical dialog.

27. The non-transitory computer-readable medium of claim 21, storing computer-readable instructions that, when executed by the processor, cause the device to:

simulate running the virtual assistant; and

receive from the simulated running virtual assistant one or more questions to ask the user in the hypothetical dialog.