When annotating the dataset, label each instance with its relevant subterms to preserve hierarchical structure.

When simplifying the equation, identify each subterm that can be factored out.