Identifying database records associated with a target

A method of identifying records at a database corresponding to a target determination object, the method comprising:

• accessing request data, the request data indicating a source table at the database and a target determination object;
• using stored knowledge data, generating first prompt data, the first prompt data comprising at least one prompt to be provided to a trained computerized model to cause the trained computerized model to return a set of determination paths to the target determination object;
• executing the trained computerized model using the first prompt data to generate the set of determination paths, the set of determination paths comprising a first determination path relating a source determination object corresponding to a source column of the source table and the target determination object corresponding to a target column of a target table different than the source table;
• determining a respective efficiency score for each determination path of the set of determination paths;
• determining a respective accuracy score for each determination path of the set of determination paths;
• determining that the first determination path is a best fit determination path of the set of determination paths, the determining that the first determination path is the best fit determination path being based on the respective efficiency scores for each determination path of the set of determination paths and the respective accuracy scores for each determination path of the set of determination paths;
• and
• determining a set of records from the source table that are associated with the target determination object using the first determination path.

A non-transitory machine-readable medium comprising instructions thereon that, when executed by at least one processor, causes the at least one processor to perform operations comprising:

• accessing request data, the request data indicating a source table at a database and a target determination object;
• using stored knowledge data, generating first prompt data, the first prompt data comprising at least one prompt to be provided to a trained computerized model to cause the trained computerized model to return a set of determination paths to the target determination object;
• executing the trained computerized model using the first prompt data to generate the set of determination paths, the set of determination paths comprising a first determination path relating a source determination object corresponding to a source column of the source table and the target determination object corresponding to a target column of a target table different than the source table;
• determining a respective efficiency score for each determination path of the set of determination paths;
• determining a respective accuracy score for each determination path of the set of determination paths;
• determining that the first determination path is a best fit determination path of the set of determination paths, the determining that the first determination path is the best fit determination path being based on the respective efficiency scores for each determination path of the set of determination paths and the respective accuracy scores for each determination path of the set of determination paths;
• and
• determining a set of records from the source table that are associated with the target determination object using the first determination path.