This paper contributes to the field of measuring (un)linkability in communication systems; a subproblem of privacy protection. We propose an attacker state model for attacks on unlinkability of partial identities named linkability graph. It covers probabilistic linkability attacks based on heterogeneous and time-variant characteristics. From our model, we derive linkability measures and argue prospects for safeguard design. Our model reduces space and time complexity compared to other contributions in literature. This enables simulative privacy analysis of complex context-aware systems that employ multiple partial identities per user.