History-Conditioned Joint-Prefix Alignment for Generative Recommendation

arXiv:2608.29179 · cs.IR, cs.AI · Submitted 2026-08-29 · Read on arXiv

cs.IR, cs.AI

Submitted: 2026-08-29

Updated: 2026-09-26

License: http://creativecommons.org/licenses/by/4.0/

The gist: Generative recommendation encodes items as hierarchical semantic identifiers (SIDs) and retrieves the next item through autoregressive decoding.

Terminology

Calibration — 31× in this paper · explained in 7 episodes
Alignment — 22× in this paper · explained in 1 episode
Pruning — 21× in this paper · explained in 4 episodes
Generative Recommendation — 11× in this paper · explained in 2 episodes
Quantization — 7× in this paper · explained in 12 episodes
Benchmark — 6× in this paper · explained in 2 episodes
Ablation — 3× in this paper · explained in 1 episode
Conference on Knowledge Discovery and Data Mining — 3× in this paper
Aggregation — 2× in this paper · explained in 1 episode
Distillation — 2× in this paper · explained in 2 episodes
In International Conference on Learning Representations — 2× in this paper
International Conference on Information and Knowledge Management — 2× in this paper
Pointwise Mutual Information (PMI) — 2× in this paper · explained in 3 episodes
Representation Learning — 2× in this paper · explained in 2 episodes
Tokenization — 2× in this paper · explained in 2 episodes
Transformer — 2× in this paper · explained in 2 episodes
Adam — 1× in this paper · explained in 2 episodes
Conference on Web Search and Data Mining — 1× in this paper
Consistency — 1× in this paper · explained in 3 episodes
Contrastive Learning — 1× in this paper · explained in 5 episodes
Decode — 1× in this paper · explained in 2 episodes
Generalization — 1× in this paper · explained in 13 episodes
Grounding — 1× in this paper · explained in 3 episodes
International Conference on World Wide Web — 1× in this paper
Knowledge Distillation — 1× in this paper · explained in 11 episodes

Abstract

Generative recommendation encodes items as hierarchical semantic identifiers (SIDs) and retrieves the next item through autoregressive decoding. Standard next-token prediction, however, does not explicitly cover the multimodal transitions present in interaction sequences, leaving the ground-truth SID vulnerable to irreversible pruning at early beam-search branches. Across three public benchmarks, we find that 91.9%--96.6% of retrieval failures occur within the first two decoding steps. We therefore propose Temporal Autoregressive Alignment (TAAL). During training, TAAL constructs a joint (c 1,c 2) soft target from historical transitions and aligns the early-prefix distribution with a forward KL objective. During inference, it calibrates candidate scores with pointwise mutual information (PMI) to reduce the influence of globally frequent prefixes. On Amazon Beauty, Instruments, and Yelp, TAAL improves NDCG@10 over the standard baseline by 39.5%, 6.7%, and 28.6%, respectively, while increasing full-SID survival by 3.9%--16.6%. Beam-width analysis further shows that the relative survival gain grows as the beam narrows, reaching 39.4% at B=5.

History-Conditioned Joint-Prefix Alignment for Generative Recommendation

Terminology

Abstract

Sources

Related papers