Systems and methods for automatically generating source code

Patent metadata

Field	Value
Publication number	US-2024362209-A1
Application number	US-202318362143-A
Country	US
Kind code	A1
Filing date	Jul 31, 2023
Priority date	Apr 27, 2023
Publication date	Oct 31, 2024
Grant date	—

Abstract

A computer-implemented method is disclosed. The method includes: receiving a request for retrieval of data satisfying one or more criteria, the request including at least one data request parameter; searching a database storing example queries based on the request to identify at least one matching query; providing, to a large language model (LLM), an input prompt to generate a query purporting to retrieve data satisfying the one or more criteria, the input prompt including the at least one data request parameter and the at least one matching query as an example; and receiving, from the LLM, a result including the generated query.

First claim

1 . A computer-implemented method, comprising:
receiving a request for retrieval of data satisfying one or more criteria, the request including at least one data request parameter; searching a database storing example queries based on the request to identify at least one matching query; providing, to a large language model (LLM), an input prompt to generate a query purporting to retrieve data satisfying the one or more criteria, the input prompt including the at least one data request parameter and the at least one matching query as an example; and receiving, from the LLM, a result including the generated query.

2 . The method of claim 1 , further comprising:
generating a first embedding based on the request in an embedding space comprising embeddings of one or more previous queries; identifying a matching embedding for the first embedding in the embedding space; and retrieving a second query associated with the matching embedding.

3 . The method of claim 2 , wherein generating the first embedding comprises generating an embedding based on the at least one data request parameter.

4 . The method of claim 2 , wherein identifying the matching embedding comprises performing a search of the embedding space to locate an embedding associated with a successful previous query that is closest to the first embedding.

5 . The method of claim 1 , further comprising:
transmitting, to an endpoint, the generated query; receiving, from the endpoint, a response indicating an error associated with the generated query; and providing, to the LLM, a further input prompt for generating a revised query, the further input prompt including error data associated with the error.

6 . The method of claim 5 , wherein the error data includes an indication of one or more error locations in computer code associated with the generated query.

7 . The method of claim 6 , wherein an error location is identified by at least one of a line number or character number containing erroneous code.

8 . The method of claim 1 , further comprising:
transmitting, to an endpoint, the generated query; receiving, from the endpoint, a response indicating that the generated query is accepted by the endpoint; and updating the database by including the generated query.

9 . The method of claim 8 , wherein the response includes first data from the endpoint associated with the at least one data request parameter.

10 . The method of claim 2 , wherein identifying the matching embedding comprises determining that a second embedding satisfies a distance criterion with respect to the first embedding, and wherein the second query comprises a query, retrieved from cache memory, that is associated with the second embedding.

11 . A computing system, comprising:
a processor; a memory coupled to the processor, the memory storing processor-executable instructions that, when executed by the processor, are to cause the processor to:
receive a request for retrieval of data satisfying one or more criteria, the request including at least one data request parameter;
search a database storing example queries based on the request to identify at least one matching query;
provide, to a large language model (LLM), an input prompt to generate a query purporting to retrieve data satisfying the one or more criteria, the input prompt including the at least one data request parameter and the at least one matching query as an example; and
receive, from the LLM, a result including the generated query.

12 . The computing system of claim 11 , wherein the instructions, when executed, are to further cause the processor to:
generate a first embedding based on the request in an embedding space comprising embeddings of one or more previous queries; identify a matching embedding for the first embedding in the embedding space; and retrieve a second query associated with the matching embedding.

13 . The computing system of claim 12 , wherein generating the first embedding comprises generating an embedding based on the at least one data request parameter.

14 . The computing system of claim 12 , wherein identifying the matching embedding comprises performing a search of the embedding space to locate an embedding associated with a successful previous query that is closest to the first embedding.

15 . The computing system of claim 11 , wherein the instructions, when executed, are to further cause the processor to:
transmit, to an endpoint, the generated query; receive, from the endpoint, a response indicating an error associated with the generated query; and provide, to the LLM, a further input prompt for generating a revised query, the further input prompt including error data associated with the error.

16 . The computing system of claim 15 , wherein the error data includes an indication of one or more error locations in computer code associated with the generated query.

17 . The computing system of claim 16 , wherein an error location is identified by at least one of a line number or character number containing erroneous code.

18 . The computing system of claim 11 , wherein the instructions, when executed, are to further cause the processor to:
transmit, to an endpoint, the generated query; receive, from the endpoint, a response indicating that the generated query is accepted by the endpoint; and update the database by including the generated query.

19 . The computing system of claim 11 , wherein the response includes first data from the endpoint associated with the at least one data request parameter.

20 . A non-transitory processor-readable medium storing processor-executable instructions that, when executed by a processor, are to cause the processor to:
receive a request for retrieval of data satisfying one or more criteria, the request including at least one data request parameter; search a database storing example queries based on the request to identify at least one matching query; provide, to a large language model (LLM), an input prompt to generate a query purporting to retrieve data satisfying the one or more criteria, the input prompt including the at least one data request parameter and the at least one matching query as an example; and receive, from the LLM, a result including the generated query.

Assignees

• Shopify Inc

Inventors

• Almaer Dion
• Geddes David
• Hwang Teddy
• Gai Darius

Classifications

• G06F16/245

  Query processing · CPC title

• G06F16/2433 Primary

  Query languages · CPC title

• G06F8/30 Primary

  Creation or generation of source code · CPC title