System architecture for interactive query processing

A system comprising:

• a non-transitory memory storing instructions executable to process a query;
• and
• a processor configured to execute the instructions and to:
• receive a first query from a user device;
• communicate the first query to artificial intelligence (AI) engine circuitry, wherein the AI engine circuitry executes a pre-trained natural language processing model on the first query, the pre-trained natural language processing model having been trained with a plurality of terminologies and a plurality of primary queries;
• receive, from the AI engine circuitry, a first parsed query response, the first parsed query response including at least one of a first intent in the first query or a first entity in the first query;
• determine a first application of a plurality of applications associated with the at least one of the first entity or the first intent based on a plurality of decision trees and a plurality of mappings, wherein the plurality of decision trees provides pre-programmed actions responsive to the at least one of the first entity or the first intent, and the plurality of mappings maps the at least one of the first entity or the first intent to the first application;
• determine a first application query associated with the at least one of the first entity or the first intent based on the plurality of decision trees and the plurality of mappings, wherein the plurality of mappings maps the at least one of the first entity or the first intent to the first application query;
• communicate the first application query to the first application;
• receive first result information from the first application responsive to the first application query;
• communicate the first result information to the user device;
• receive a second query from the user device, wherein the second query is subsequent to the first query and a context of the second query is selected from the first result information;
• communicate the second query to the AI engine circuitry;
• receive, from the AI engine circuitry, a second parsed query response, the second parsed query response including a second intent in the second query;
• determine that an action corresponding to the second intent requires a corresponding entity field;
• determine that the second parsed query response did not include a second entity;
• when it is determined that the second parsed query response did not include the second entity, review the first query to determine a first context of the first query and use the determined first context as a current context of the second query, wherein the first context includes the first entity;
• determine a second application associated with the second intent and the first entity;
• determine a second application query associated with the second intent and the first entity;
• communicate the second application query to the second application;
• receive second result information from the second application responsive to the second application query, wherein the second application provides the second result information that is more detailed than the first result information provided by the first application;
• and
• communicate the second result information to the user device.

A method comprising:

• in artificial intelligence (AI) engine circuitry:
• training a natural language processing model with training data to generate a pre-trained natural language processing model, the training data comprising:
• a plurality of terminologies and corresponding intents,
• a plurality of primary queries and corresponding intents, and
• a plurality of conversation flows;
• receiving a first query from query processing hardware;
• receiving a second query from the query processing hardware, wherein the second query is subsequent to the first query and a context of the second query is selected from first result information;
• processing the first query with the pre-trained natural language processing model to parse the first query into a first parsed query response including at least one of a first intent of the first query or a first entity in the first query;
• processing the second query with the pre-trained natural language processing model to parse the second query into a second parsed query response including a second intent in the second query;
• and
• communicating the first parsed query response and the second parsed query response to the query processing hardware;
• in the query processing hardware:
• receiving the first parsed query response;
• determining a first application query associated with the at least one of the first entity or the first intent based on a plurality of decision trees and a plurality of mappings, wherein the plurality of decision trees provides pre-programmed actions responsive to the at least one of the first entity or the first intent, and the plurality of mappings maps the at least one of the first entity or the first intent to the first application query;
• querying a first application with the first application query;
• receiving the first result information from the first application;
• receiving the second parsed query response;
• determining that an action corresponding to the second intent requires a corresponding entity field;
• determining that the second parsed query response did not include a second entity;
• when it is determined that the second parsed query response did not include the second entity, reviewing the first query to determine a first context of the first query and using the determined first context as a current context of the second query, wherein the first context includes the first entity;
• determining a second application associated with the second intent and the first entity;
• determining a second application query associated with the second intent and the first entity;
• communicating the second application query to the second application;
• receiving second result information from the second application responsive to the second application query, wherein the second application provides the second result information that is more detailed than the first result information provided by the first application;
• and
• communicating the second result information to the query processing hardware.