Mobile Edge Computing (MEC) stands as an indispensable technology in the facilitation of 5G networks, enabling the deployment of widely-used services on edge servers situated in close proximity to consumer electronics. Within the MEC framework, a central role is attributed to edge service composition (ESC), pivotal in bolstering functionality and ameliorating user experiences. Presently, prevailing methods for ESC predominantly center on the prioritization of Quality of Service (QoS) optimization, presenting a solitary optimal composite service for consumer electronics invocation. Regrettably, this approach sidelines the significance of solution diversity within composite services, potentially resulting in service overload and the suboptimal utilization of edge resources. To surmount these challenges, this study integrates digital twin (DT) technology and diversified search mechanisms into the MEC domain, offering an innovative diversified top-k service composition methodology known as DSC-DT. By harnessing the capabilities of DT technology, DSC-DT enables the emulation and assessment of diverse composite service solutions within a virtual space. Specifically, the proposed methodology models the procedure of service composition within a DT environment as an issue of subgraph isomorphism. This is succeeded by the configuration of the diversification process as an independent set predicament within an undirected graph, efficiently resolved through a greedy algorithmic paradigm. It is noteworthy that DSC-DT accommodates a gamut of query types, including normal queries, constraint queries, and optimal queries. The efficacy and efficiency of the proposed approach are corroborated through comprehensive experiments conducted upon authentic datasets.