Scalable multi-service virtual assistant platform using machine learning
The present invention is a masterbot architecture in a scalable multi-service virtual assistant platform that can construct a fluid and dynamic dialogue by assembling responses to end user utterances from two kinds of agents, information agents and action agents. A plurality of information agents obtain at least one information value from a parsed user input and/or contextual data. A plurality of action agents perform one or more actions in response to the parsed user input, the contextual data, and/or the information value. A masterbot arbitrates an activation of the plurality of information agents and the plurality of action agents. The masterbot comprises access to a machine-learning module to select an appropriate action agent, where one or more information agents are activated based on the selected appropriate action agent.
1. A system for a virtual assistant platform, comprising:
at least one memory device that stores computer-executable instructions; and
at least one processor configured to access the memory device, wherein the processor is configured to execute the computer-executable instructions to implement:
a plurality of information agents to obtain at least one information value from a parsed user input and/or contextual data from one or more contextual data sources, wherein the parsed user input is generated from one or more utterances received from a user;
a plurality of action agents to perform one or more actions in response to the parsed user input, the contextual data, and/or the at least one information value; and
at least one masterbot to activate one or more of the plurality of information agents and the plurality of action agents, the masterbot comprising access to a machine-learning module to select an appropriate action agent, wherein one or more information agents are activated based on the selected appropriate action agent.
2. The system of claim 1 , wherein the one or more information agents are selected using the machine learning module.
3. The system of claim 1 , wherein at least one set-value connection interrelates a given action agent and a given information agent via at least one set-value action, wherein the set-value action for the given action agent is used to set an information value for the given information agent which is later used by the given information agent.
4. The system of claim 1 , wherein at least one inference connection interrelates two information agents, wherein the inference connection interrelates the two information agents amongst themselves which determine when a given information agent activates one or more related information agents in order to fulfill one or more information values of the given information agent which are later used by the given information agent.
5. The system of claim 1 , wherein at least one prerequisite connection interrelates at least one information agent and at least one action agent, and wherein the prerequisite connection activates one or more prerequisite information agents to fulfill one or more prerequisite conditions for a given action agent.
6. The system of claim 1 , wherein at least one of the plurality of information agents is adapted to understand the parsed user input from the user to extract the information value directly from the parsed user input.
7. The system of claim 1 , wherein at least one of the plurality of information agents is adapted to infer the information value from the parsed user input, wherein the inferring is activated in response to determining that a given information agent cannot obtain the information value by understanding the parsed user input directly.
8. The system of claim 7 , wherein at least one of the plurality of information agents is adapted to infer the information value by accessing one or more contextual data sources for the contextual data and is adapted to infer the information value from the parsed user input and the contextual data.
9. The system of claim 8 , wherein at least one of the plurality of information agents is adapted to query the user for the information value by generating a natural language query using a natural language generation (NLG) module, wherein the querying is activated in response to determining that the information agent cannot obtain the information value by understanding the parsed user input or inferring the information value.
10. The system of claim 9 , wherein at least one of the plurality of information agents is adapted to obtain the information value by activating one or more other information agents, wherein the one or more other information agents are activated in response to determining that the at least one information agent cannot obtain the information value by understanding the parsed user input or inferring the information value.
11. The system of claim 1 , wherein at least one of the plurality of action agents performs the one or more actions by triggering one or more internal services to perform one or more services.
12. The system of claim 11 , wherein at least one of the plurality of action agents performs the one or more actions by accessing one or more Application Programming Interface (API) gateways to external services to perform the one or more services.
13. The system of claim 1 , wherein the system activates one or more of the information agents in order to meet one or more prerequisite connections for a given action agent, wherein the given action agent utilizes the one or more activated information agents to obtain one or more information values needed to perform the one or more actions.
14. The system of claim 1 , wherein at least one of the plurality of action agents is adapted to generate a natural language reply to the user using a natural language generation (NLG) module.
15. The system of claim 1 , wherein at least one of the plurality of action agents is adapted to change a system state.
16. The system of claim 1 , wherein the masterbot evaluates a performance of the plurality of information agents and the plurality of action agents using the machine learning module to adjust the activation of the information agents and the action agents.
17. The system of claim 1 , wherein the masterbot generates a response to the user using a natural language generation (NLG) module from information supplied by the information agents and/or the action agents.
18. A non-transitory, computer-readable medium storing program code for a multi-service virtual assistant platform which, when executed by a processor, cause the processor to perform computer-executable operations to:
obtain at least one information value from a parsed user input and/or contextual data from one or more contextual data sources using a plurality of information agents, wherein the parsed user input is generated from one or more utterances received from a user;
perform one or more actions in response to the parsed user input, the contextual data, and/or the at least one information value using a plurality of action agents; and
activate one of the plurality of information agents and the plurality of action agents by selecting an appropriate action agent using a machine-learning module, wherein one or more information agents are activated based on the selected appropriate action agent.
19. A computer-implemented method for a multi-service virtual assistant platform, comprising:
obtaining at least one information value from a parsed user input and/or contextual data from one or more contextual data sources using a plurality of information agents, wherein the parsed user input is generated from one or more utterances received from a user;
performing one or more actions in response to the parsed user input, the contextual data, and/or the at least one information value using a plurality of action agents; and
activating one of the plurality of information agents and the plurality of action agents by selecting an appropriate action agent using a machine-learning module, wherein one or more information agents are activated based on the selected appropriate action agent.
20. The computer-implemented method of claim 19 , wherein the one or more information agents are selected using the machine learning module.