Generative-AI SQL-generation based on natural language input
AssigneeAbcs Insights Inc.
InventorsJerome Shimizu, Michael Sheng
Life of this patent
Estimating the term…
Abstract
Systems and methods are disclosed for generating customized advertising audiences using artificial intelligence. The system analyzes a request for a desired audience, input through a natural language chat interface or the like, and automatically converts that input into structured traits and a database query based on the structured traits. A large language model (LLM) is used to analyze the request to extract meaningful characteristics (like age, behavior, or shopping patterns) and generate a structured query language (SQL) command to identify matching users in a database. The system can then deliver content to the matching users users or update marketing strategies based on performance.
Claims20
A method performed by one or more hardware processors, the method comprising:
- receiving an audience request describing attributes of purchasing behavior of individuals of a desired target audience;
- analyzing, using a large language model (LLM), the audience request to generate one or more structured audience characteristics that define identifiable traits, behaviors, or criteria associated with the desired target audience;
- generating, using the same LLM that generated the one or more structured audience characteristics, a structured query language (SQL) query based on the one or more structured audience characteristics, the LLM generating the SQL query that includes syntactically valid executable SQL code;
- executing the SQL query that includes the syntactically valid executable SQL code generated by the same LLM that generated the one or more structured audience characteristics and the structured SQL query, the executing of the SQL query causing an advertiser system to apply the structured audience characteristics against a user database comprising user identifiers to identify a subset of users corresponding to the desired target audience;
- and
- executing, by the same LLM that generated the one or more structured audience characteristics, generated the structured SQL query, and executed the SQL query, an ad campaign deployment system causing delivery of a digital advertising campaign to the identified subset.
A system comprising:
- at least one processor;
- and
- at least one memory component storing instructions that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:
- receiving an audience request describing attributes of purchasing behavior of individuals of a desired target audience;
- analyzing, using a large language model (LLM), the audience request to generate one or more structured audience characteristics that define identifiable traits, behaviors, or criteria associated with the desired target audience;
- generating, using the same LLM that generated the one or more structured audience characteristics, a structured query language (SQL) query based on the one or more structured audience characteristics, the LLM generating the SQL query that includes syntactically valid executable SQL code;
- executing the SQL query that includes the syntactically valid executable SQL code generated by the same LLM that generated the one or more structured audience characteristics and the structured SQL query, the executing of the SQL query causing an advertiser system to apply the structured audience characteristics against a user database comprising user identifiers to identify a subset of users corresponding to the desired target audience;
- and
- executing, by the same LLM that generated the one or more structured audience characteristics, generated the structured SQL query, and executed the SQL query, an ad campaign deployment system causing delivery of a digital advertising campaign to the identified subset.
A non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor to perform operations comprising:
- receiving an audience request describing attributes of purchasing behavior of individuals of a desired target audience;
- analyzing, using a large language model (LLM), the audience request to generate one or more structured audience characteristics that define identifiable traits, behaviors, or criteria associated with the desired target audience;
- generating, using the same LLM that generated the one or more structured audience characteristics, a structured query language (SQL) query based on the one or more structured audience characteristics, the LLM generating the SQL query that includes syntactically valid executable SQL code;
- executing the SQL query that includes the syntactically valid executable SQL code generated by the same LLM that generated the one or more structured audience characteristics and the structured SQL query, the executing of the SQL query causing an advertiser system to apply the structured audience characteristics against a user database comprising user identifiers to identify a subset of users corresponding to the desired target audience;
- and
- executing, by the same LLM that generated the one or more structured audience characteristics, generated the structured SQL query, and executed the SQL query, an ad campaign deployment system causing delivery of a digital advertising campaign to the identified subset.
Dependencies are inferred from the claim text.