Techniques for automated messaging
Techniques for automated messaging are described. Some embodiments are particularly directed to techniques for automated business-to-consumer messaging leveraging social-networking information. In one embodiment, an apparatus may comprise a user profile component operative to retrieve a user profile associated with a user account for a messaging service and a messaging automation component operative to receive a messaging initiation from a messaging endpoint on a client device; and configure an automated response based on the user profile in response to the messaging initiation. Other embodiments are described and claimed.
1. A computer-implemented method, comprising:
receiving a user message from a user;
submitting the user message to a natural language processor;
receiving, from the natural language processor, one or more user intent keywords indicating an intent of the user;
accessing a response message repository;
retrieving, from the response messaging repository, a response message based on a user intent keyword mapping between the response message and the one or more user intent keywords; and
sending the retrieved response message in response to the user message.
2. The method of claim 1 wherein the response message repository comprises a plurality of response messages and a user intent keyword mapping from each response message to one or more user intent keywords.
3. The method of claim 1 , wherein the response message repository is initialized by:
retrieving all or a portion of the response messages in an initially-configured response message repository;
submitting the retrieved response messages to the natural language processor;
receiving, from the natural language processor, a user intent keyword mapping for each of the submitted response messages defining an association between the each submitted response message and one or more user intent keywords; and
storing the user intent keyword mappings in the response message repository associated with their respective response messages.
4. The method of claim 2 further comprising:
receiving an administrator-generated response message in response to the user message;
submitting the user message and the administrator-generated response message to the natural language processor;
receiving, from the natural language processor, a keyword mapping defining an association between the administrator-generated response message and one or more additional user intent keywords; and
adding the administrator-generated response message and the one or more additional user intent keywords to the response message repository.
5. The method of claim 4 further comprising:
receiving a confirmation that the one or more additional user intent keywords were accurately associated with the administrator-generated response message; and
refining the training of the natural language processor by providing feedback of the accuracy of the user intent keyword mapping between the administrator-generated response message and the one or more additional user intent keywords.
6. The method of claim 1 , the response message defining one or more dynamic fields, further comprising:
customizing the response message by providing values for the one or more dynamic fields.
7. The method of claim 1 further comprising:
receiving, from the natural language processor, one or more parameters defining details of the intent of the user.
8. The method of claim 1 further comprising:
receiving, from the natural language processor, a confidence associated with each returned user intent keyword;
wherein a particular response message is retrieved from the response message repository only if the confidence associated with the user intent keyword for the particular response message exceeds a predetermined threshold.
9. A system, comprising:
a processor; and
software that, when executed by the processor, causes the system to:
receive a user message from a user;
submit the user message to a natural language processor;
receive, from the natural language processor, one or more user intent keywords indicating an intent of the user;
access a response message repository;
retrieve, from the response messaging repository, a response message based on a user intent keyword mapping between the response message and the one or more user intent keywords; and
send the retrieved response message in response to the user message.
10. The system of claim 9 wherein the response message repository comprises a plurality of response messages and a user intent keyword mapping from each response message to one or more user intent keywords, the software further causing the system to:
receive an administrator-generated response message in response to the user message;
submit the user message and the administrator-generated response message to the natural language processor;
receive, from the natural language processor, a keyword mapping defining an association between the administrator-generated response message and one or more additional user intent keywords; and
add the administrator-generated response message and the one or more additional user intent keywords to the response message repository.
11. The system, of claim 9 , wherein the response message repository is initialized by software which further causes the system to:
retrieve all or a portion of the response messages in an initially-configured response message repository;
submit the retrieved response messages to the natural language processor;
receive, from the natural language processor, a user intent keyword mapping for each of the submitted response messages defining an association between the each submitted response message and one or more user intent keywords; and
store the user intent keyword mappings in the response message repository associated with their respective response messages.
12. The system of claim 10 , the software further causing the system to:
receive a confirmation that the one or more additional user intent keywords were accurately associated with the administrator-generated response message; and
refine the training of the natural language processor by providing feedback of the accuracy of the keyword mapping between the administrator-generated response message and the one or more additional user intent keywords.
13. The system of claim 9 , the software further causing the system to:
receive, from the natural language processor, one or more parameters defining details of the intent of the user.
14. The system of claim 9 , the software further causing the system to:
receive, from the natural language processor, a confidence associated with each returned user intent keyword;
wherein a particular response message is retrieved from the response message repository only if the confidence associated with the user intent keyword for the particular response message exceeds a predetermined threshold.
15. A non-transitory, computer-readable medium storing software that, when executed, cause a system to:
receive a user message from a user;
submit the user message to a natural language processor;
receive, from the natural language processor, one or more user intent keywords indicating an intent of the user;
access a response message repository;
retrieve, from the response messaging repository, a response message based on a user intent keyword mapping between the response message and the one or more user intent keywords; and
send the retrieved response message in response to the user message.
16. The medium of claim 15 wherein the response message repository comprises a plurality of response messages and an intent keyword mapping from each response message to one or more user intent keywords, the software further causing the system to:
receive an administrator-generated response message in response to the user message;
submit the user message and the administrator-generated response message to the natural language processor;
receive, from the natural language processor, a keyword mapping defining an association between the administrator-generated response message and one or more additional user intent keywords; and
add the administrator-generated response message and the one or more additional user intent keywords to the response message repository.
17. The medium of claim 15 , wherein the response message repository is initialized by software which further causes the system to:
retrieve all or a portion of the response messages in an initially-configured response message repository;
submit the retrieved response messages to the natural language processor;
receive, from the natural language processor, a user intent keyword mapping for each of the submitted response messages defining an association between the each submitted response message and one or more user intent keywords; and
store the user intent keyword mappings in the response message repository associated with their respective response messages.
18. The medium of claim 16 , the software further causing the system to:
receive a confirmation that the one or more additional user intent keywords were accurately associated with the administrator-generated response message; and
refine the training of the natural language processor by providing feedback of the accuracy of the user intent keyword mapping between the administrator-generated response message and the one or more additional user intent keywords.
19. The medium of claim 15 , the software further causing the system to:
receive, from the natural language processor, one or more parameters defining details of the intent of the user.
20. The medium of claim 15 , the software further causing the system to:
receive, from the natural language processor, a confidence associated with each returned user intent keyword;
wherein a particular response message is retrieved from the response message repository only if the confidence associated with the user intent keyword for the particular response message exceeds a predetermined threshold.