Pith. sign in

REVIEW 4 major objections 5 minor 23 references

PushDualGen: Enabling LLMs to Generate Semantic IDs with Interpretable Copy for Industrial Push Recommendation

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read PushDualGen claims that generating the semantic ID first, then the push copy as a skippable explanation, yields both accurate and interpretable push recommendations at industrial scale, with online A/B gains of 8.50% in effective play…

desk verdict Real deployment, interesting design order, but the A/B test doesn't isolate the copy mechanism, so the headline causal claim is under-supported. read the letter →

arxiv 2608.07989 v1 pith:PZWVPLPW submitted 2026-08-08 cs.IR

Manjia Lin , Da Li , Yan Wang , Yong Jin , Zheming Ding , Wei Yuan , Lei Yan , Yanan Xia
show 4 more authors
Lu Zhang Fan Yang Xuanping Li Yanan Niu
This is my paper · ORCID
classification cs.IR
keywords pushnotificationgenerativerecommendationsemanticIDinterpretablecopyparallelonlineA/BtestingcontentecosystemLLM-based
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

PushDualGen is trying to establish that generative recommendation can be made practical for push notifications by reversing the generation order used in reasoning-augmented systems: first decide the item, encoded as a semantic ID (SID), then optionally write the notification copy as a skippable explanation of that decision. The paper reports a 14-day A/B test over roughly 150 million users in which the deployed system raised effective play rate by 8.50% relative and cut dissatisfaction rate by 37.70% relative, at roughly 100K queries per second. A sympathetic reader should care because the design claims to remove both obstacles that have kept LLM-based generative recommendation out of push: the black-box intractability of SID outputs and the prohibitive inference cost of chain-of-thought reasoning. The same mechanism is also claimed to improve the content ecosystem by exposing long-tail videos.

What carries the argument

The load-bearing object is the Parallel Semantic ID: each video is mapped to $M = 8$ parallel embeddings, and each embedding slot is quantized by nearest-centroid assignment against its own $K = 512$ codebook, producing an 8-token discrete ID that compresses long interaction histories into a form an LLM can read and emit. Because the slots are parallel, deeper SID tokens do not accumulate the error that plagues residual quantization. The second mechanism is the generation order: the model first outputs the SID tokens, then a separator token $\langle id2text\_sep \rangle$, then the copy tokens, with a weighted copy loss; the copy is 'skippable' because at serving time it can be dropped without any extra model pass. The third mechanism is representation fusion: the Top-$N$ generated SIDs are embedded into $e_s$ and combined with user features as $e'_u = \alpha e_u + \beta e_s$, which then drives approximate nearest-neighbor retrieval over video embeddings. Multi-token binding, which merges frequent $n$-grams and SIDs into single tokens, is what keeps the input length and training cost manageable.

What would settle it

Run two additional A/B arms—one with SID ranking but copy generation disabled, and one with the baseline ranking but PushDualGen's generated copy—and check whether each reproduces the 8.50% effective-play-rate gain and 37.70% dislike-rate drop; if the copy-only arm matches the full-system results, the central claim about SID-first generation is falsified.

Watch

Extended reading notes

Core claim

The central claim is that one small generator can both predict which video a user will engage with and write the copy that justifies the notification, provided the SID is generated before the copy and the copy is treated as optional. The paper builds SIDs from eight parallel embedding slots, each independently quantized with K-means into 512-token codebooks, so no token inherits the error of earlier tokens as residual quantization does. After a paired text-to-SID and SID-to-text adaptation phase, the generator consumes compressed user histories and emits SID tokens followed by a separator and copy tokens; at serving time the copy can be skipped entirely, and the generated SIDs are fused into the user representation used for ANN retrieval. The paper's evidence is the production deployment and the online A/B results, alongside an ablation showing that each component—parallel SIDs, SID adaptation, token freezing, and multi-token binding—improves SID prediction accuracy over the raw backbone.

Load-bearing premise

The A/B test compares the entire PushDualGen system against an unspecified cascade pipeline, with no variant isolating the copy text from the SID-based ranking, so the headline gains could partly or wholly come from the copy alone or from a different candidate pool rather than from the SID-first generation design.

Editorial extensions

If this is right

  • If the A/B gains hold, SID-first generation becomes a template for latency-sensitive generative recommendation: interpretability no longer requires paying the decoding cost of a chain of thought.
  • The 37.70% relative drop in dissatisfaction rate implies the copy shown to users is better matched to the video actually served, reducing the post-click mismatch that push notifications are especially prone to.
  • The exposure shift toward torso and tail content means the generative SID signal generalizes beyond historically popular items, which is the mechanism behind the claimed cold-start benefit.
  • Because the SID is a discrete, loggable code, operators can trace which generated item the user was offered while discarding the copy, giving an audit trail that pure free-text recommendations do not provide.
  • The frozen-vocabulary adaptation and multi-token binding allow the approach to be retrained incrementally on daily click logs, which is what makes deployment at 100K QPS and billion-user scale feasible.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A test the paper does not run would separate the two levers: a SID-only arm (ranking changed, no copy) and a copy-only arm (baseline ranking, new copy) would tell us whether the 8.50% play-rate gain and 37.70% dislike drop come from better item selection or from better wording—the current A/B conflates them.
  • The copy is a self-explanation: it is generated by the same model that produced the SID, so it cannot serve as independent evidence of why the item was chosen; an audit study comparing generated copy to human-judged video content would probe whether the explanation is faithful or merely plausible.
  • Because SIDs are K-means centroids over parallel embeddings, they drift when the embedding model is updated; monitoring SID-to-text alignment over time would be a natural safeguard, and the paper does not report on codebook stability.
  • If the efficiency claim transfers, the same SID-first-with-skippable-copy ordering could be applied outside push—for example, to feed recommendation or assistant-style explanations—wherever a discrete decision code and an optional natural-language rationale are both needed.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. PushDualGen is a generative push-recommendation framework for Kuaishou. It encodes videos into parallel semantic IDs (SIDs) via K-means quantization of multi-slot embeddings, aligns SIDs with an LLM through Text2SID/SID2Text objectives, fine-tunes the LLM to generate a SID followed by a skippable push copy, and fuses the generated Top-N SIDs with user features for ANN-based retrieval. The paper reports deployment at roughly 100K QPS, a 14-day A/B test on ~150M users against an undescribed cascading-pipeline baseline, and claims relative gains of +8.50% in effective play rate and -37.70% in dissatisfaction rate, plus improved long-tail exposure. Offline ablations report SID-prediction Pass@k for several component variants, and a qualitative case study is used to support the interpretability claim.

Significance. If the central claim holds, PushDualGen is a notable industrial data point: a lightweight, SID-first generative recommender with skippable copy that improves both engagement and post-click satisfaction in a deployed push system serving a billion users. The paper's strengths include the large-scale online A/B test with CUPED adjustment, the concrete deployment details, and the plausibility of parallel SIDs as a cheaper alternative to chain-of-thought for generative recommendation. However, the causal attribution of the reported gains to the SID-first/copy design is not established by the evidence presented, and the interpretability claim is supported only by self-generated copy examples. The significance is therefore conditional on additional experimental isolation and transparency.

major comments (4)
  1. [Section 4.1, Table 1] The A/B test compares the full PushDualGen system with an 'online service based on a cascading pipeline' that is never described. The treatment arm simultaneously changes the candidate-retrieval mechanism (Eq. (7) fusion followed by ANN search), the push copy text (Section 3.2), and the exposure distribution across head/torso/tail content (Section 5). With no online arm that fixes the copy while varying the SID ranking, or vice versa, the abstract's causal attribution—that the 'SID-first, skippable-copy' design yields +8.50% effective play rate and -37.70% dissatisfaction rate—is not supported. The offline ablation in Table 2 measures only SID-prediction Pass@k and does not decompose copy quality or post-click satisfaction. Please provide at least a high-level description of the baseline pipeline and either online component-level ablations or a clear argument for why the copy effect cannot be separated.
  2. [Table 1 and Section 4.1] The paper reports only point estimates and a p<0.05 marker. There are no confidence intervals, no effect-size distributions, and no details on the CUPED adjustment beyond the citation. The relative changes are very heterogeneous across metrics (Click PV +0.43%, DAU +0.05%, Eff. Play Rate +8.50%, Dis. Rate -37.70%), and the two engagement metrics are small in relative terms. Without confidence intervals or a description of the variance reduction achieved by CUPED, the reader cannot judge whether the 'significant' framing is robust or whether the headline gains are driven by a few segments. Please report CUPED-adjusted confidence intervals and, if possible, absolute-level metrics for both arms.
  3. [Section 3.2 and Section 6] The interpretability claim is load-bearing for the paper's motivation, but it is not validated. The copy is generated by the same LLM that generated the SID, after conditioning on that SID; therefore the copy is self-consistent with the SID by construction, and the qualitative examples in Figure 3 only show that the model can produce text that looks related to the user's history. There is no human evaluation, no measurement of whether the copy changes users' understanding of why the video was recommended, and no evidence that the copy corresponds to the actual retrieval logic rather than to a plausible post-hoc text. Please provide a user study, a faithfulness metric against a held-out explanation, or an explicit limitation statement that the copy is an interpretation aid rather than a verified explanation.
  4. [Section 4.3 and Eq. (7)] The offline ablation evaluates only SID prediction (Pass@k) and does not evaluate copy generation at all, even though copy generation is one of the two outputs of the generator and one of the claimed reasons for the online gains. Additionally, the fusion hyperparameters α=β=1 and Top-N=20 are presented as fixed choices with no sensitivity analysis; the contribution of the representation-fusion component is therefore untested. Since Eq. (7) is central to online serving, reporting Pass@k with different N, α, and β values would help establish that the reported gains are not driven by an arbitrary fusion setting.
minor comments (5)
  1. [Section 3.1] The notation for the number of codebooks is inconsistent: the text and Eq. (1) use M slots, while Eq. (4) and the surrounding text use L for the codebook index. Please unify the notation, and also check that the summation index in Eq. (3) is consistent with the target token definition.
  2. [Abstract and References] The paper refers to 'OneRec-Thinking' in the abstract and 'OneRec-Think' in Section 2.1 and the references; please use a consistent name throughout.
  3. [Table 1 and Section 5] The exposure distribution analysis in Figure 2 reports proportions without error bars, significance tests, or a description of how the 'Online Service' exposure is computed. The claim that PushDualGen 'optimizes the content ecosystem' would be stronger with confidence intervals or a time-series analysis.
  4. [Section 3.1 and Appendix A] The term 'skippable copy' is never formally defined. It should be clear whether the copy is always generated at inference, whether it is always displayed, or whether it can be omitted based on latency or user state; this affects the interpretation of the online results.
  5. [Section 4.1] The sentence 'we conduct incremental training based on user click logs with approximately 3.6B tokens every day' is ambiguous about whether this is the total training-token volume per day or the daily accumulation of user logs used for a single training run. Please clarify.

Circularity Check

1 steps flagged · score 2.0 of 10

No significant circularity in the main A/B evidence; the only self-consistency issue is the post-hoc 'interpretable copy', which is not independently validated.

  1. self definitional [Section 3.2 (Eq. 5) and Section 6 Case Study]
    "PushDualGen retains the ability to generate copy then SIDs. This helps us to understand what is contained within SIDs, as well as the recommendation logic behind PushDualGen."

    The copy used as the 'explanation' is generated by the same model that generated the SID: Eq. (5) conditions the copy on the predicted SID. Since the SID and the copy come from the same generator, the copy is self-consistent with the SID by construction and cannot independently reveal the model's recommendation logic. The paper does not validate the copy against user comprehension or an external explanation ground truth, so the interpretability claim reduces to the model paraphrasing its own output.

full rationale

The central effectiveness claim rests on a 14-day online A/B test (Section 4.1, Table 1) against a live cascading-pipeline service. This is an external benchmark, not an output of the model, so the main claim is not circular. The offline Pass@k evaluation (Table 2) also compares model-generated SIDs to held-out ground-truth SIDs, so it is a genuine predictive test. The parallel SID design borrows from prior work by overlapping authors (CREM, CoMa), but the paper specifies the training objective in Appendix A and tests the design in an ablation (w/o Parallel SID), so this self-citation is not load-bearing. The one self-consistency issue is the 'interpretable copy': the copy is generated by the same model that generates the SID (Eq. 5), and Section 6 presents it as evidence of the recommendation logic, but it is a post-hoc output of the model itself and is not validated against an independent measure. This affects the interpretability sub-claim, not the A/B effectiveness result, so overall circularity is low.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The ledger shows that the system relies on a small number of hand-chosen structural parameters (M=8 slots, K=512 centroids, N=20 top SIDs, fusion weights set to 1) plus a set of unverified domain assumptions. The most consequential assumption is that the K-means-quantized parallel embeddings from a CREM-style training produce SIDs that transfer cleanly to LLM generation and to the online ANN search. No new physical entities are introduced; the components are built from existing techniques.

free parameters (6)
  • Number of parallel SID slots M = 8
    Chosen by hand in Sec 3.1; each video is encoded into 8 parallel embeddings, and each is quantized into one token.
  • Codebook size K per slot = 512
    Chosen in Sec 3.1 and Appendix A for K-means; larger or smaller K would change SID granularity and model capacity.
  • Top-N SIDs for fusion = 20
    Set in Sec 3.3; the top 20 generated SIDs are embedded and fused into the user representation.
  • Fusion weights alpha and beta = 1, 1
    Eq (7). The paper states they are set to 1 in applications, giving equal weight to the generative SID signal and user features.
  • Copy loss weight lambda_Copy = not specified
    Eq (6). Balances SID and copy objectives; the value is not given in the text.
  • Number of compression tokens = 8
    Implementation Details; used to map multimodal video content into 8 latent embeddings before quantization.
assumptions (5)
  • domain assumption The parallel embedding model trained with InfoNCE and language modeling (CREM/CoMa paradigm) produces embeddings whose K-means clusters yield stable, meaningful SIDs.
    Required by Eq (1) and Appendix A; if the embeddings do not capture video semantics, SIDs are noise and generation cannot work.
  • domain assumption Fine-tuning only the SID token embeddings (freezing the rest of the vocabulary) is sufficient for the LLM to learn SID semantics.
    Sec 3.1; the ablations show only a small drop without token freeze, so this assumption is not strongly validated.
  • domain assumption The LLM's SID prediction task (Pass@k) is a valid proxy for push recommendation quality.
    Sec 4.3 uses Pass@k as the primary offline metric; the paper does not show a correlation between Pass@k and the online metrics.
  • ad hoc to paper The copy generated after the SID is a faithful, interpretable explanation of the recommendation.
    Sec 3.2 and Sec 6; the copy is produced by the same model and has no independent validation.
  • domain assumption The online A/B experiment and the CUPED adjustment correctly isolate the treatment effect of PushDualGen.
    Sec 4.1; the baseline 'cascading pipeline' is not described and no confidence intervals are given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PushDualGen: Enabling LLMs to Generate Semantic IDs with Interpretable Copy for Industrial Push Recommendation." pith.science (2026). https://pith.science/paper/PZWVPLPW

@misc{pith2026260807989,
  author       = {Pith},
  title        = {Pith review of: PushDualGen: Enabling LLMs to Generate Semantic IDs with Interpretable Copy for Industrial Push Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PZWVPLPW}},
  note         = {Machine review of arXiv:2608.07989}
}
read the original abstract

Push recommendation in KuaiShou proactively delivers personalized content to nearly one billion users to facilitate their engagement. Recently, generative recommendation has achieved end-to-end user personalization through semantic ID. However, their black- box characteristics make recommendation logics difficult to trace, hindering their deployment. OneRec-Thinking addresses this by incorporating CoT before generating SIDs, but this significantly increases inference cost. To support large-scale industrial applications, we propose PushDualGen, a lightweight generator, which first generates the SID and then produces a copy as a skippable explanation. PushDualGen has been deployed in Kuaishou's push recommendation system. Online A/B tests demonstrate the effectiveness of PushDualGen, delivering significant improvements in both user attraction and satisfaction. The effective play rate for videos recommended to users has relatively increased by 8.50%, while the dissatisfaction rate has relatively fallen by 37.70%. In the long term, PushDualGen optimises the content ecosystem, providing more exposure for long-tail videos.

Figures

Figures reproduced from arXiv: 2608.07989 by the authors.

Figure 1
Figure 1. Overview of PushDualGen. Module (A) compresses videos into Parallel SID and aligns them with the token of LLMs [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The proportion of exposure among different groups. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. It is clear that PushDualGen captures the users’ prefer [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Parallel embedding model and SID construction. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 11 canonical work pages

  1. [1]

    Alex Deng, Ya Xu, Ron Kohavi, and Toby Walker. 2013. Improving the Sen- sitivity of Online Controlled Experiments by Utilizing Pre-Experiment Data. InProceedings of the Sixth ACM International Conference on Web Search and Data Mining (WSDM ’13). Association for Computing Machinery, 123–132. http://doi.org/10.1145/2433396.2433413

    arXiv 2013

  2. [2]

    Jiaxin Deng, Shuo Wang, Kang Cai, Long Ren, Qigen Hu, Qiang Luo, Qian Wang, Wei Ding, and Guorui Zhou. 2025. OneRec: Unifying Retrieve and Rank with Generative Recommender and Iterative Preference Alignment.arXiv preprint arXiv:2502.18965(2025). http://doi.org/10.48550/arXiv.2502.18965

    arXiv doi:10.48550/arxiv.2502.18965 2025

  3. [3]

    Yan Gao, Viral Gupta, Jinyun Yan, Changji Shi, Zhongen Tao, PJ Xiao, Curtis Wang, Shipeng Yu, Rómer Rosales, Ajith Muralidharan, and Shaunak Chatterjee

    work page

  4. [4]

    Yupeng Hou, Jiacheng Li, Ashley Shin, Jinsung Jeon, Abhishek Santhanam, Wei Shao, Kaveh Hassani, Ning Yao, and Julian McAuley. 2025. Generating Long Semantic IDs in Parallel for Recommendation. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD ’25). Association for Computing Machinery, Toronto, ON, Canada, 956...

    work page arXiv 2025

  5. [5]

    Huxiao Ji, Haitao Yang, Linchuan Li, Shunyu Zhang, Cunyi Zhang, Xuanping Li, and Wenwu Ou. 2024. TIM: Temporal Interaction Model in Notification System. InProceedings of the 2024 International Conference on Multimedia Retrieval (ICMR ’24). Association for Computing Machinery, Phuket, Thailand, 1120–1124. http://doi.org/10.1145/3652583.3657614

    work page arXiv 2024

  6. [6]

    Da Li, Yuxiao Luo, Keping Bi, Jiafeng Guo, Wei Yuan, Biao Yang, Yan Wang, Fan Yang, Tingting Gao, and Guorui Zhou. 2026. Compressing then Matching: An Efficient Pre-training Paradigm for Multimodal Embedding.arXiv preprint arXiv:2511.08480(2026). http://doi.org/10.48550/arXiv.2511.08480

    arXiv doi:10.48550/arxiv.2511.08480 2026

  7. [7]

    Xinyu Lin, Chaoqun Yang, Wenjie Wang, Yongqi Wang, Fuli Feng, and Tat-Seng Chua. 2025. Efficient Inference for Large Language Model-based Generative Recommendation. InThe Thirteenth International Conference on Learning Repre- sentations (ICLR 2025). https://openreview.net/forum?id=VdrACDjN40

    work page 2025

  8. [8]

    Lihao Liu, Yan Wang, Biao Yang, Da Li, Jiangxia Cao, Yuxiao Luo, Xiang Chen, Xiangyu Wu, Wei Yuan, Fan Yang, Guiguang Ding, Tingting Gao, and Guorui Zhou. 2026. CREM: Compression-Driven Representation Enhancement for Mul- timodal Retrieval and Comprehension.arXiv preprint arXiv:2602.19091(2026). http://doi.org/10.48550/arXiv.2602.19091

    work page doi:10.48550/arxiv.2602.19091 2026

Show all 23 references
  1. [9]

    Zhanyu Liu, Jiaxin Deng, Junda She, Kuo Cai, Lejian Ren, Qigen Hu, Qiang Luo, Shiyao Wang, Xinchen Luo, Xingmei Wang, Yi Su, Yunfan Wu, Zixing Zhang, and Guorui Zhou. 2025. OneRec-Think: In-Text Reasoning for Generative Recommendation.arXiv preprint arXiv:2510.11639(2025). htt...

    2025 doi

  2. [10]

    Zhao Liu, Yichen Zhu, Yiqing Yang, Guoping Tang, Rui Huang, Qiang Luo, Xiao Lv, Ruiming Tang, Kun Gai, and Guorui Zhou. 2025. DiffGRM: Diffusion-based Generative Recommendation Model.arXiv preprint arXiv:2510.21805(2025). http://doi.org/10.48550/arXiv.2510.21805

    2025 doi

  3. [11]

    Conor O’Brien, Huasen Wu, Shaodan Zhai, Dalin Guo, Wenzhe Shi, and Jonathan J. Hunt. 2022. Should I Send This Notification? Optimizing Push Notifi- cations Decision Making by Modeling the Future.arXiv preprint arXiv:2202.08812 (2022). http://doi.org/10.48550/arXiv.2202.08812

    2022 doi arXiv

  4. [12]

    Tran, Jonah Samost, Maciej Kula, Ed H

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Kesha- van, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Q. Tran, Jonah Samost, Maciej Kula, Ed H. Chi, and Maheswaran Sathiamoorthy. 2023. Recommender Systems with Generative Retrieval. InAdvances in Neural ...

    2023

  5. [13]

    RecGPT Team. 2025. RecGPT Technical Report.arXiv preprint arXiv:2507.22879 (2025). http://doi.org/10.48550/arXiv.2507.22879

    2025 doi arXiv

  6. [14]

    RecGPT Team. 2025. RecGPT-V2 Technical Report.arXiv preprint arXiv:2512.14503 (2025). http://doi.org/10.48550/arXiv.2512.14503

    2025 doi

  7. [15]

    Teng Shi, Chenglei Shen, Weijie Yu, Shen Nie, Chongxuan Li, Xiao Zhang, Ming He, Yan Han, and Jun Xu. 2025. LLaDA-Rec: Discrete Diffusion for Parallel Seman- tic ID Generation in Generative Recommendation.arXiv preprint arXiv:2511.06254 (2025). http://doi.org/10.48550/arXiv.2511.06254

    2025 doi

  8. [16]

    Chengbing Wang, Yang Zhang, Zhicheng Wang, Tianhao Shi, Keqin Bao, Fuli Feng, and Tat-Seng Chua. 2025. Decoding in Latent Spaces for Efficient Inference in LLM-based Recommendation.arXiv preprint arXiv:2509.11524(2025). http: //doi.org/10.48550/arXiv.2509.11524

    2025 doi

  9. [17]

    Jin Xu, Zhifang Guo, Jinzheng He, Hangrui Hu, Ting He, Shuai Bai, Keqin Chen, Jialin Wang, Yang Fan, Kai Dang, Bin Zhang, Xiong Wang, Yunfei Chu, and Junyang Lin. 2025. Qwen2.5-Omni Technical Report. arXiv:2503.20215 [cs.CL] https://arxiv.org/abs/2503.20215

    2025 arXiv

  10. [18]

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jia...

    2025 arXiv

  11. [19]

    Bo Zhao, Koichiro Narita, Burkay Orten, and John Egan. 2018. Notification Volume Control and Optimization System at Pinterest. InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD ’18). Association for Computing Machinery, Lon...

    2018

  12. [20]

    Guorui Zhou, Honghui Bao, Jiaming Huang, Jiaxin Deng, Jinghao Zhang, Junda She, Kuo Cai, Lejian Ren, Lu Ren, Qiang Luo, Qianqian Wang, Qigen Hu, Rongzhou Zhang, Ruiming Tang, Shiyao Wang, Wuchao Li, Xiangyu Wu, Xinchen Luo, Xingmei Wang, Yifei Hu, Yunfan Wu, Zhanyu Liu, Zhiyan...

  13. [21]

    +” indicates visible tokens and “-

    Guorui Zhou, Hengrui Hu, Hongtao Cheng, Huanjie Wang, Jiaxin Deng, Jinghao Zhang, Kuo Cai, Lejian Ren, Lu Ren, Liao Yu, Pengfei Zheng, Qiang Luo, Qianqian Wang, Qigen Hu, Rui Huang, Ruiming Tang, Shiyao Wang, Shujie Yang, Tao Wu, Wuchao Li, Xinchen Luo, Xingmei Wang, Yi Su, Yu...

    2025 doi arXiv

  14. [2018]

    InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD ’18)

    Near Real-Time Optimization of Activity-Based Notifications. InProceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD ’18). Association for Computing Machinery, London, United Kingdom, 283–292. http://doi.org/10.1145/3219819.3219880

  15. [2026]

    http://doi.org/10.48550/arXiv.2512.24762

    OpenOneRec Technical Report.arXiv preprint arXiv:2512.24762(2026). http://doi.org/10.48550/arXiv.2512.24762

    2026 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.