26_08_08_sprig_cikm_accepted

Our paper “SPRIG: Semantic-ID-enhanced Paths for Knowledge Graph-based Generative Recommendation” has been accepted at ACM CIKM 2026 🎉

SPRIG is a generative recommender that integrates Semantic IDs—discrete codes derived from item content such as text and audio—into knowledge-graph path reasoning, grounding recommendations in structured item relationships while achieving competitive performance with fewer parameters and lower compute cost.