| CPC G06F 16/235 (2019.01) [G06F 16/215 (2019.01); G06F 16/2358 (2019.01)] | 20 Claims |
|
1. A computer-implemented method comprising:
storing a set of policy constraints of an organization;
generating a plurality of datasets and a set of associations between datasets of the plurality of datasets and policy constraints of the set of policy constraints, comprising, repeating for each of a plurality of natural language data analytics requests:
receiving, by an online system, a natural language data analytics request based on data stored in one or more data stores,
generating an execution plan for processing the natural language data analytics request using a machine learning based language model, the execution plan comprising a set of steps, each step comprising a set of instructions, wherein one or more steps of the execution plan are associated with a particular policy constraint of the set of policy constraints,
generating a dataset by executing the execution plan, and
storing an association between the dataset and the particular policy constraint;
receiving a request to modify a policy constraint;
responsive to receiving the request to modify the policy constraint, regenerating execution plans, comprising:
identifying, using the set of associations between the datasets and the policy constraints, a subset of datasets associated with the policy constraint being modified; and
for each dataset from the subset of datasets:
identifying the execution plan used to generate the dataset,
identifying a subset of steps of the execution plan associated with the policy constraint being modified, and
regenerating instructions of the subset of steps.
|