Generative Recommendation Models Struggle with New Items, Study Finds

Researchers have explored the limitations of Semantic-ID (SID)-based generative recommendation systems, particularly their ability to handle "cold items"—new items with unseen semantic tokens. A temporal analysis revealed that while these models can sometimes predict future items based on observed tokens and prefixes, they falter when encountering entirely new atomic tokens or weakly supported SID paths. The study frames SID generation as a hierarchical semantic bucketing process, suggesting that current methods are compositional but not fully open-ended, pointing towards future research in more independent SID spaces and dynamic textual contexts. AI

IMPACT Highlights limitations in generative recommendation systems for handling new content, suggesting areas for future research in token independence and contextualization.

RANK_REASON Academic paper published on arXiv detailing limitations of generative recommendation systems. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• alphaXiv
• arXiv
• CatalyzeX
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• Influence Flower
• ScienceCast
• Semantic ID

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →