Method and Apparatus for Live Chat Integration

• US 20130346886A1
• Filed: 06/26/2013
• Published: 12/26/2013
• Est. Priority Date: 06/26/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Cauchois, Callixte, Lebrun, Alexandre
• Granted Patent: US 9,973,457 B2
• Time in Patent Office: Days
• US Class Current: 715/758
• CPC Class Codes: H04L 51/046 Interoperability with other network applications or services

Abstract

A Live Chat (LC) system, apparatus, and method combine virtual agent (VA) and LC in a manner enabling a user to receive resolution of a contact request at a reduced cost and enhancing both the user and LC operator'"'"'s experience. VAs may source response content for easier or more frequent user queries, whereas VAs may transfer a user query to an LC operator whenever a user query becomes too complicated, enabling the LC to source the response content. From the user'"'"'s point of view, escalation from the VA to the LC operator happens in the very same user screen view used to contact the VA, enhancing the user'"'"'s experience by providing seamless escalation. Conversation history between the user and the VA or formatted data related thereto may be provided to the LC operator, expediting the LC operator'"'"'s response time and reducing a number of interactions needed to resolve the user'"'"'s query.

Claims

1. A method for implementing a live chat system, the method comprising:

providing a response to a query, entered by a user via a user screen view on a user interface and submitted to a virtual agent server hosting at least one virtual agent, the response provided to the user from the virtual agent server via the user screen view, in a manner maintaining continuity of conversation in the user screen view independent of whether the at least one virtual agent or a live chat operator, having access to the at least one virtual agent via a live chat server operatively coupled to the virtual agent, sources content of the response.

2. The method of claim 1, further comprising:

providing a conversation history, between the user and the at least one virtual agent, to the live chat operator via an operator screen view on an operator interface in a manner enabling the live chat operator to source the response within a context of the conversation history.

3. The method of claim 2, wherein providing the conversation history includes (i) mirroring all query and response interactions since a last was response sourced by the live chat operator in the operator screen view or (ii) gathering metadata based on the user query, formatting the metadata gathered, and providing the metadata formatted in the operator screen view, wherein the metadata gathered includes at least one of:

a user name, an identification of a root problem triggering the query, a support center contacted by the user, a product of interest to the user, other contextual information related to the query, or a combination thereof.

4. The method of claim 2, further comprising:

identifying a start of a chat session based on an initial query entered by the user;

allocating a unique global session identifier for the chat session started;

providing the unique global session identifier with the conversation history provided; and

identifying an end of the chat session based on an end query entered by the user and, responsively, deallocating the unique global session identifier allocated.

5. The method of claim 1, further comprising:

applying at least one rule to the query to trigger a transfer of the query from the at least one virtual agent to the live chat operator to enable the live chat operator to process the query and source the response, wherein the at least one virtual agent sources the content of the response if the transfer is not triggered.

6. The method of claim 5, wherein applying the at least one rule includes processing a linguistic analysis of the query using a knowledge database operatively coupled to the virtual agent server and triggering the transfer if a result of the linguistic analysis indicates that the query is outside of a scope of the knowledge database.

7. The method of claim 5, wherein applying the at least one rule include a user property, wherein the user property includes a very important person (VIP) priority status and further wherein the transfer is triggered if an identifier of a user profile associated with the user indicates the VIP priority status.

8. The method of claim 5, further comprising:

gathering metadata related to query and response interaction with the user at the virtual agent independent of whether or not the at least one virtual agent or the live chat operator sources the content of the response;

recording each query and response interaction at the virtual agent independent of whether or not the at least one virtual agent or the live chat operator sources the content of the response, each query and response interaction optionally recorded in association with the metadata gathered; and

utilizing a response sourced by the live chat operator in response to a user query transferred from the at least one virtual agent to the live chat operator to enhance a knowledge database operatively coupled to the virtual agent server, enabling the at least one virtual agent to source the response to a future occurrence of the user query and preventing future transfer of the user query to the live chat operator.

9. The method of claim 1, further comprising:

applying a linguistic analysis to the query at the at least one virtual agent and determining at least one proposed response to the query and a confidence metric associated with the at least one proposed response determined;

transferring the query with the at least one proposed response and the confidence metric determined to the live chat operator via an operator screen view on an operator interface; and

enabling the live chat operator to override the at least one proposed response transferred or to select from the at least one proposed response transferred to source the response in the user screen view.

10. The method of claim 9, further including:

determining if the response from the live chat operator is sourced to the user screen view within a predetermined timeframe; and

sourcing the response from the at least one virtual agent if the predetermined timeframe is exhausted, wherein the at least one virtual agent sources the content of the response based on the at least one proposed response and a comparison of the confidence metric determined with a confidence threshold.

11. The method of claim 10, wherein if the confidence metric is above the confidence threshold, the response sourced by the at least one virtual agent is selected from the at least one proposed response and wherein if the confidence metric is not above the confidence threshold, the response sourced by the at least one virtual agent is an alternative response.

12. A live chat system comprising:

a virtual agent server, hosting at least one virtual agent accessible by a user via a user screen view on a user interface, the virtual agent server operatively coupled to a live chat server, the at least one virtual agent being accessible to a live chat operator via an operator screen view on an operator interface, the virtual agent server including a user side network interface and an operator side network interface, the at least one virtual agent configured to;

provide a response to a query, entered by a user via the user screen view on a user interface and submitted to the user side network interface of the virtual agent server, the response provided via the user side network interface to the user from the virtual agent server via the user screen view, in a manner maintaining continuity of conversation in the user screen view independent of whether the at least one virtual agent or the live chat operator sources content of the response.

13. The live chat system of claim 12, wherein the at least one virtual agent is configured to provide a conversation history, between the user and the at least one virtual agent, to the live chat server via the operator side network interface, in a manner enabling the live operator to source the response within a context of the conversation history.

14. The live chat system of claim 13, wherein the at least one virtual agent includes a dialog manager, wherein, to provide the conversation history, the dialog manager is further configured to:

(i) mirror all query and response interactions since a last response was sourced by the live chat operator via the operator screen view or (ii) gather metadata based on the user query, format the metadata gathered, and provide the metadata formatted to the live chat server via the operator side network interface, wherein the metadata gathered includes at least one of;

a user name, an identification of a root problem triggering the query, a support center contacted by the user, a product of interest to the user, other contextual information related to the query, or a combination thereof.

15. The live chat system of claim 12, wherein the at least one virtual agent includes a dialog manager and a natural language understanding module (NLU) and is operatively coupled to a knowledge database, and wherein the at least one virtual agent is further configured to:

apply at least one rule to the query to trigger a transfer of the query to the live chat operator to enable the live chat operator to process the query and source the response, wherein the at least one virtual agent sources the content of the response if the transfer is not triggered, and wherein, to apply the at least one rule, the NLU is configured to perform a linguistic analysis of the query using the knowledge database and the dialog manager is further configured to trigger the transfer if a result of the linguistic analysis indicates that the query is outside of a scope of the knowledge database.

16. The live chat system of claim 15, wherein the at least one rule includes a user property including a very important person (VIP) priority status and further wherein the dialog manager is further configured to trigger the transfer if an identifier of a user profile associated with the user indicates the VIP priority status.

17. The live chat system of claim 12, wherein the at least one virtual agent is operatively coupled to a knowledge database, the at least one virtual agent further configured to:

gather metadata related to query and response interaction with the user;

record each query and response interaction independent of whether or not the at least one virtual agent or the live chat operator sources the content of the response, each query and response interaction optionally recorded in association with the metadata gathered; and

utilize a response sourced by the live chat operator in response to a user query transferred from the at least one virtual agent to the live chat operator to enhance the knowledge database, enabling the at least one virtual agent to source the response to a future occurrence of the user query and preventing future transfer of the user query to the live chat operator.

18. The live chat system of claim 12, wherein the at least one virtual agent is further configured to:

identify a start of a chat session based on an initial query entered by the user;

allocate a unique global session identifier for the chat session started;

provide the unique global session identifier with the conversation history provided; and

identify an end of the chat session based on an end query entered by the user and, responsively, deallocate the unique global session identifier allocated.

19. The live chat system of claim 12, wherein the at least one virtual agent includes an NLU configured to apply a linguistic analysis to the query and to determine at least one proposed response to the query and a confidence metric associated with the at least one proposed response determined, and further wherein the at least one virtual agent includes a dialog manager configured to:

transfer the query, with the at least one proposed response and the confidence metric determined, to the live chat operator in the operator screen view enabling the live chat operator to override the at least one proposed response transferred or to select from the at least one proposed response transferred to source the response in the user screen view;

determine if the response from the live chat operator is sourced to the user screen view within a predetermined timeframe; and

source the response if the predetermined timeframe is exhausted, wherein the dialog manager sources the content of the response based on the at least one proposed response and a comparison of the confidence metric determined by the NLU with a confidence threshold, wherein if the confidence metric is above the confidence threshold, the response sourced is selected by the dialog manager from the at least one proposed response and if the confidence metric is not above the confidence threshold, the response sourced is an alternative response.

20. A non-transitory computer-readable medium having stored thereon a sequence of instructions which, when loaded and executed by a processor, causes the processor to:

provide a response to a query, entered by a user via a user screen view and submitted to a virtual agent server hosting at least one virtual agent, the response provided to the user from the virtual agent server via the user screen view, in a manner maintaining continuity of conversation in the user screen view independent of whether the at least one virtual agent or a live chat operator, having access to the at least one virtual agent via a live chat server, sources content of the response.