The present disclosure provides a method, which may comprise obtaining a plurality of essay prompts, processing each essay prompt of said plurality of essay prompts using a natural language processing (NLP) algorithm to extract at least one feature from said each essay prompt and, using said at least one feature from said each essay prompt, generating two or more subsets of said plurality of essay prompts, wherein an essay prompt in a subset satisfies a measure of similarity with respect to each other essay prompt in said subset such that a single essay can be responsive to each essay prompt in said subset.