In LTL Model Checking while converting the LTL formula to an equivalent ω-automaton, we can neglect the fairness constraint in acceptance conditions for positive occurrences of [ϕ U ψ] we can neglect the fairness constraint for negative occurrences of [ϕ U ψ].

What do we mean by positive and negative occurrences here?