IoT Device Management (DM) refers to the remote administration of customer devices to ensure better Quality of Experience (QoE). In practice, DM tends to be ensured by multiple actors such as opera- tors or device manufacturers, each operating independently through their legacy DM solution. Such siloed DM capabilities are limited in addressing IoT threats related to device dependencies, such as cascading failures since they are not aware of the global topology of dependencies. Deter- mining this topology is challenging, requiring to infer dependencies from the data held by different actors. In this work, we propose an easy-to- use framework that infers such a topology on demand by accessing and aggregating data from legacy DM solutions. Thanks to Semantic Web standards, the framework enables unified data extraction, interpretation, and usage across heterogeneous DM solutions to support decision-making in business applications such as customer care. We leverage the digital twin technology to expose on-demand the global topology of dependencies. We integrated our solution within the in-use Orange’s digital twin platform Thing in the future and demonstrate its effectiveness by au- tomatically inferring dependencies in a smart home scenario managed by ground truth DM solutions such as Orange’s implementation of USP Controller and Samsung’s SmartThings Platform.