| CPC G06F 16/9032 (2019.01) [G06N 20/00 (2019.01)] | 20 Claims |
|
1. A method, comprising:
receiving, at a computer system and from a user system, an inquiry into data stored in a database accessible by the computer system using a database query code format, wherein the inquiry is not in the database query code format;
generating a set of descriptors for a sequence of two or more queries corresponding to the inquiry, wherein generating the set of descriptors includes executing, by a processor of the computer system, a text generative machine learning model trained using analysis history data selected based at least in part on performance characteristics for previous query execution using the database;
executing, by the computer system, a sequence-to-sequence machine learning model in a first direction to produce corresponding database query code for the set of descriptors, wherein the sequence-to-sequence machine learning model is trained using query history data selected based at least in part on performance characteristics for previous query execution using the database;
executing, by the computer system, the sequence-to-sequence machine learning model again, in a reverse direction as compared to the first direction, to generate one or more alternative query code submodules corresponding to the set of descriptors;
transmitting annotated queries, generated based on the one or more alternative query code submodules, to the user system; and
submitting, in response to receiving input from the user system for the annotated queries, at least a portion of the corresponding database query code to a database server for execution using the database.
|