Metraipolis › archive › dc_f2b0cbdee927

STAR: Structured Tokenization and Target-Aware Interest Representation for PCVR Prediction — arXiv:2608.12986

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.12986
License (as archived)
CC0-1.0
Archived
2026-10-19T04:55:14.423Z
SHA-256 of the archived text
26799d3dbc1715bf167a41674bdb4ba199f2d928523b347ead734ed077360777
Stable URL of this page
https://metraipolis.com/documento/dc_f2b0cbdee927

How to cite

STAR: Structured Tokenization and Target-Aware Interest Representation for PCVR Prediction — arXiv:2608.12986 — arXiv (en), https://arxiv.org/abs/2608.12986 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f2b0cbdee927 (Metraipolis archive dc_f2b0cbdee927, sha256 26799d3dbc1715bf167a41674bdb4ba199f2d928523b347ead734ed077360777)
BibTeX
@misc{dcf2b0cbdee927,
  title = {STAR: Structured Tokenization and Target-Aware Interest Representation for PCVR Prediction — arXiv:2608.12986},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.12986},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f2b0cbdee927; sha256 26799d3dbc1715bf167a41674bdb4ba199f2d928523b347ead734ed077360777},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.12986. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Post-click conversion rate (PCVR) prediction is a core ranking task in industrial recommender systems. Modern ranking models must jointly capture heterogeneous non-sequential features, multi-behavior user sequences, and target-item-aware user interests, while remaining robust to high-cardinality sparse features, missing values, and train-inference inconsistencies. In this paper, we present STAR (Structured Tokenization and Target-Aware Interest Representation), a practical framework for the KDD Cup 2026 Tencent UniRec Challenge. STAR combines structured feature tokenization with target-aware interest representation on top of a HyFormer-style multi-sequence backbone. It introduces high-cardinality signal recovery, explicit user-item interaction tokens, target-aware sequence decoding, and a weighted user-item contrastive auxiliary objective inspired by InfoNCE. We further align the training and inference pipelines by reconstructing feature remapping tables and structural hyperparameters from the saved training configuration. Experiments on the challenge dataset identify the components that most reliably improve ranking AUC, while LogLoss is reported as a calibration diagnostic. The main ablation study shows a large gain from temporal context, with smaller but useful contributions from contrastive alignment, target-aware interest encoding, and high-cardinality sequence feature recovery.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-08-14 in cs.IR).
Categorie arXiv: Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2608.12986 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T04:46:09.756Z dalla pagina https://arxiv.org/catchup/cs.IR/2026-08-14?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).