The accurate transfer of discrete particle information to continuum Eulerian fields, known as mapping or coarse-graining, plays a critical role in unresolved CFD-DEM modeling, governing both numerical stability and physical fidelity. Over the years, a variety of strategies have been proposed, spanning from local methods such as the satellite point scheme to non-local approaches based on kernel functions, diffusion, or hybrid formulations. Each method balances trade-offs between smoothness, conservation, computational efficiency, and applicability to complex grid configurations or non-spherical particles. This perspective summarizes the general methodology, representative implementations, and typical applications of existing mapping algorithms, and analyzes their respective merits and limitations. Particular attention is given to challenges associated with small grid-to-particle size ratios, irregular geometries, computational costs, and multi-physics coupling. Emerging directions, including adaptive and hybrid schemes, consistency with turbulence modeling, extensions to polydisperse and non-spherical particles, and machine learning-aided mapping acceleration, are discussed. Continued efforts in these areas promise to improve the robustness, accuracy, and scalability of CFD-DEM simulations, ultimately enabling more generalized and reliable modeling of complex multiphase flows in both research and industrial applications.