Reputation and trust-based models have gained popularity recently because they have been shown to be promising in the area of trust management. Despite this fact, building reliable systems still remains a challenge. Proposed models focus on historical and online information to determine the reputation of domain members. However, the dynamic nature of reputation and trust requires an equally dynamic approach to computing and resolving trust related issues in any domain. This paper proposes a reliable and novel dynamic framework that utilises a data-driven approach for trust management. The framework uses past interactions, recent and anticipated future trust values of every identity in the domain. The proposed framework is critically evaluated and compared with existing work through experiments. The advantage of this proactive framework compared to other approaches is that informed decisions about the domain can be made before misbehaviour occurs.