There is no hard requirement for how much context should be (optionally) provided for disambiguation. In practice, the context provided could be a sentence, a paragraph, or even an entire article. While it is not required that the context contain the entity text being grounded, in practice, context usually consists of text surrounding and including the entity. Overall, given that disambiguation models integrated from Adeft as well as the models made available independently by Gilda rely on word frequency-derived features, it is generally helpful if context text contains terms characteristic to the given sense of the ambiguous string. Additionally, since Adeft models have the ability to recognize defining patterns (i.e., an explicit spelled out definition of an ambiguous acronym), any context text containing such patterns is useful.

Species prioritization is applied in case matches are found to proteins from multiple species (human, mouse, yeast, etc.). In case of matches to multiple species for a given input, this (optional) input list is used to prioritize matches. The priority list can be determined based on the user case or knowledge about the context in which the given entity text appeared. The species list is adopted from UniProt's popular organisms list.