2026. Vision-Language Foundation Models with Hierarchical Motion Memory for Long Video Question Answering and Event Localization. Journal of Data Intelligence and AI Systems. 1, 1 (May 2026).