Origin-destination (OD) flow forecasting supports urban mobility decisions by predicting the volume and patterns of human movement across a city. A new paper on arXiv introduces SeoulMMOD, a large-scale multimodal benchmark designed for this task.
The paper's abstract argues that most existing public OD benchmarks provide only a partial view of citywide mobility. SeoulMMOD appears intended to fill that gap, though the abstract does not specify the exact data sources, modalities, or geographic scope included in the benchmark.
The significance of the work lies in offering a more complete basis for evaluating forecasting models. As with any new benchmark, its practical value will depend on the details presented in the full paper, which are not available in the abstract alone.