Named Data Networking (NDN) adopts onedimensional hierarchical naming and lookup operations based on longest prefix matching. On the other hand, users may express their requests with multi-dimensional attributes in order to meet the intrinsic spatial- temporal nature of the data existing in IoT applications. To bridge the gap between multi- dimensional user requests and one-dimensional hierarchical names in NDN applications, a novel middleware supporting multi-dimensional naming is designed to support efficient Interest expressions through name translation and optimization, as well as keep original NDN operations unchanged. Simulation experiments show that the proposed solution can improve the content retrieval efficiency and outperform the other baseline schemes in terms of the number of Interest packets and content retrieval time.