Chatbot artificial intelligence
A system and method simulate a chat-based conversation with a human user. The system and method receive a text message through the chat system that enables a receiver to transmit an automatically generated response in real time during a chat session. The system and method recognize characters in a text message and generates an automatic prediction or classification using a conversational assistant pod that resides in a container. The system and method generate an automatic response and modify computing resources by replacing the conversational assistant pod and the container with a second conversational assistant pod and a second container without interrupting the chat session.
1. A non-transitory machine-readable medium encoded with machine-executable instructions, where execution of the machine-executable instructions is for:
receiving, by a conversational assistant pod within an on-line chat system, a free-form input from a user through a distributed network, by which the on-line chat system automatically generates a reply in response to the free-form input during a chat session;
processing the free-form input through a grammar-based natural language recognition that compares the free-form input to a vocabulary, the grammar-based natural language recognition recognizes text from the free-form input by matching the free-form input to the vocabulary;
scoring a recognized text through a machine learning model served by the conversational assistant pod that reflects a probability that the recognized text matches the free-form input from the user;
selecting a pre-generated response from a knowledge base based on the scoring of the recognized text to render the reply;
transmitting the pre-generated response to the user contemporaneously with the free-form input received from the user;
where a master node is programmed to drain traffic from the conversational assistant pod when an operating condition that precedes a failure is detected and routing further processing to a second conversational assistant pod during a session that comprises conversation with the user; and
where the second conversational assistant pod is in a same operating state as the conversational assistant pod before the conversational assistant pod fails.
2. The non-transitory machine-readable medium of claim 1 where the free-form input comprises a textual input and an image input.
3. The non-transitory machine-readable medium of claim 1 where the free-form input comprises a textual input and one or more emails.
4. The non-transitory machine-readable medium of claim 1 where the conversational assistant pod resides in a cloud.
5. The non-transitory machine-readable medium of claim 1 further comprising a container hosting the conversational assistant pod that is responsive, the conversational assistant pod, the container, and the master node, residing on a conversational cluster that is remote from and responsive to a primary cluster.
6. The non-transitory machine-readable medium of claim 5 where the primary cluster schedules a plurality of conversational clusters based on running workloads processed by the plurality of conversational clusters and the master node schedules a plurality of worker nodes comprising a plurality of containers based on an operating capacity of the plurality of worker nodes.
7. The non-transitory machine-readable medium of claim 5 where the container comprises a standalone software that includes the conversational assistant pod, a plurality of application libraries, a runtime software, a plurality of system tools, and a plurality of system libraries.
8. The non-transitory machine-readable medium of claim 1 where the machine learning model comprises a neural network.
9. The non-transitory machine-readable medium of claim 1 where the machine learning model comprises an ensemble of machine learning networks that are automatically trained via a feedback loop without a human intervention.
10. The non-transitory machine-readable medium of claim 1 where the recognized text comprises a weighted text based on an association with pre-identified intents that represent an intended purpose of a textual expression.
11. The non-transitory machine-readable medium of claim 1 further comprising mapping the scoring in metadata and associating the metadata associated with the free-form input.
12. The non-transitory machine-readable medium of claim 1 where the machine learning model comprises a multi-tree ensemble of networks joined serially that update a learning algorithm's residual errors.
13. The non-transitory machine-readable medium of claim 1 where the pre-generated response comprises a textual message and a of static image.
14. The non-transitory machine-readable medium of claim 1 where the pre-generated response comprises video responses accessible through an executable link.
15. The non-transitory machine-readable medium of claim 1 further comprising authenticating the user before initiating the chat session and tailoring the pre-generated response based on a profile associated with an authenticated user.
16. The non-transitory machine-readable medium of claim 1 further comprising a plurality of containers each hosting a plurality of conversational assistant pods that are responsive only to the master node within a conversational cluster, where the free-form input is assigned to a single conversational assistant pod through a round-robin delegation by the master node.
17. The non-transitory machine-readable medium of claim 1 where the conversational assistant pod provides for on-line insurance servicing through an on-line claims processing.
18. The non-transitory machine-readable medium of claim 1 where the conversational assistant pod provides for on-line insurance servicing through an on-line insurance quoting.
19. The non-transitory machine-readable medium of claim 1 further comprising a plurality of containers each hosting a plurality of conversational assistant pods that are responsive to the master node within a conversational cluster, where the free-form input is assigned to a single conversational assistant pod through a round-robin delegation by the master node.
20. A method comprising:
receiving, by a conversational assistant pod within an on-line chat system, a free-form input from a user through a distributed network, by which the on-line chat system automatically generates a reply in response to the free-form input during a chat session;
processing the free-form input through a grammar-based natural language recognition that compares the free-form input to a vocabulary, the grammar-based natural language recognition recognizes text from the free-form input by matching the free-form input to the vocabulary;
scoring a recognized text through a machine learning model served by the conversational assistant pod that reflects a probability that the recognized text matches the free-form input from the user;
selecting a pre-generated response from a knowledge base based on the scoring of the recognized text to render the reply; and
transmitting the pre-generated response to the user contemporaneously with the free-form input received from the user;
where a master node is programmed to drain traffic from the conversational assistant pod when an operating condition that precedes a failure is detected and routing further processing to a second conversational assistant pod during a session that comprises conversation with the user; and
where the second conversational assistant pod is in a same operating state as the conversational assistant pod before the conversational assistant pod fails.
21. The method of claim 20 where the free-form input comprises a textual input and an image input.
22. The method of claim 20 where the free-form input comprises a textual input and one or more emails.
23. The method of claim 20 where the conversational assistant pod resides in a cloud.
24. The method of claim 20 further comprising a container hosting the conversational assistant pod that is responsive the conversational assistant pod, the container, and the master node, residing on a conversational cluster that is remote from and responsive to a primary cluster.
25. The method of claim 24 where the primary cluster schedules a plurality of conversational clusters based on running workloads processed by the plurality of conversational clusters and the master node schedules a plurality of worker nodes comprising a plurality of containers based on an operating capacity of the plurality of worker nodes.
26. The method of claim 24 where the container comprises a standalone software that includes the conversational assistant pod, a plurality of application libraries, a runtime software, a plurality of system tools, and a plurality of system libraries.
27. The method of claim 20 where the machine learning model comprises a neural network.
28. The method of claim 20 where the machine learning model comprises an ensemble of machine learning networks that are automatically trained via a feedback loop without a human intervention.
29. The method of claim 20 where the recognized text comprises a weighted text based on an association with pre-identified intents that represent an intended purpose of a textual expression.
30. The method of claim 20 further comprising mapping the scoring in metadata and associating the metadata associated with the free-form input.
31. The method of claim 20 where the machine learning model comprises a multi-tree ensemble of networks joined serially that update a learning algorithm's residual errors.
32. The method of claim 20 where the pre-generated response comprises a textual message and a of static image.
33. The method of claim 20 where the pre-generated response comprises video responses accessible through an executable link.
34. The method of claim 20 further comprising authenticating the user before initiating the chat session and tailoring the pre-generated response based on a profile associated with an authenticated user.
35. The method of claim 20 further comprising a plurality of containers each hosting a plurality of conversational assistant pods that are responsive only to the master node within a conversational cluster, where the free-form input is assigned to a single conversational assistant pod through a round-robin delegation by the master node.
36. The method of claim 20 where the conversational assistant pod provides for on-line insurance servicing through an on-line claims processing.
37. The method of claim 20 where the conversational assistant pod provides for on-line insurance servicing through an on-line insurance quoting.
38. The method of claim 20 further comprising a container hosting the conversational assistant pod that is responsive to the conversational assistant pod, the container, and the master node, residing on a conversational cluster that is remote from and responsive to a primary cluster.