Overmerging occurs when a knowledge graph entity has too many records such that these records make data for the entity factually incorrect. For example, individuals with the same name, job title, educational background, and so forth, could become overmerged leading to information from two people entities included in one entity.
Knowledge graphs operate on the assumption that unknown data is not false, unlike other knowledge systems. This informs the fact that undermerging — or the lack of records within a knowledge graph entity — does not lead to factual incorrectness.