The DOI matches its Crossref record; 1 of the 1 references it cites were found in a registry or an index.
Decoding Matters: Addressing Amplification Bias and Homogeneity Issue in Recommendations for Large Language Models
Open accessAbstract
Adapting Large Language Models (LLMs) for recommendation requires careful consideration of the decoding process, given the inherent differences between generating items and natural language.Existing approaches often directly apply LLMs' original decoding methods.However, we find these methods encounter significant challenges: 1) amplification bias-where standard length normalization inflates scores for items containing tokens with generation probabilities close to 1 (termed ghost tokens), and 2) homogeneity issue-generating multiple similar or repetitive items for a user.To tackle these challenges, we introduce a new decoding approach named Debiasing-Diversifying Decoding (D 3 ).D 3 disables length normalization for ghost tokens to alleviate amplification bias, and it incorporates a text-free assistant model to encourage tokens less frequently generated by LLMs for counteracting recommendation homogeneity.Extensive experiments on real-world datasets demonstrate the method's effectiveness in enhancing accuracy and diversity.The code is available at https://github. com/SAI990323/DecodingMatters.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Each recommendation request typically requires generating a list of items, linearly scaling the cost with the list size.
“Worse yet, each recommendation request from users typically requires generating a list of items (Lin et al., 2024a; Bao et al., 2024; Zheng et al., 2024), linearly scaling the cost regarding the list size.”
From Decoding in Latent Spaces for Efficient Inference in LLM-based Recommendation (2025) · cites this paper as Bao et al., 2024
The Introduction says recommendation output should be a set of items, and §3 describes beam search generating multiple items. The paper does not state that generation cost scales linearly with list size; §3 describes generating items simultaneously rather than measuring that cost relationship. No page numbers are available in our copy. (LLM, conf=0.84)
Experimental — accuracy not yet measuredLLM-Based Generative Recommenders generate natural language recommendations without predefined items.
“LLM-Based Generative Recommenders generate natural language recommendations without predefined items, offering innovative potential (Bao et al., 2023a, 2024; Zheng et al., 2024; Hua et al., 2023; Zhu et al., 2024; Zhang et al., 2025b).”
From Decoding in Latent Spaces for Efficient Inference in LLM-based Recommendation (2025) · cites this paper as Bao et al., 2024
The Introduction describes LLM-based recommenders directly generating item representations, such as titles, and calls the approach promising. It also distinguishes recommended items from unrestricted natural-language output, while §6.1.2 refers to an item list in the dataset. The claim that recommendations are generated without predefined items is therefore not supported by this paper. (LLM, conf=0.86)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 34
- Keqin Bao, Jizhi Zhang, Xinyu Lin, Yang Zhang, Wen- jie Wang, and Fuli Feng. 2024. Large language models for recommendation: Past, present, and fu- ture. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2993-2996.
- Keqin Bao, Jizhi Zhang, Wenjie Wang, Yang Zhang, Zhengyi Yang, Yancheng Luo, Chong Chen, Fuli Feng, and Qi Tian. 2023a. A bi-step grounding paradigm for large language models in recommenda- tion systems. arXiv preprint arXiv:2308.08434. · arXiv
- Keqin Bao et al. 2023b. Tallrec: An effective and effi- cient tuning framework to align large language model with recommendation. In RecSys, pages 1007-1014. ACM.
- Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783. · arXiv
- Jesse Harte, Wouter Zorgdrager, Panos Louridas, As- terios Katsifodimos, Dietmar Jannach, and Marios Fragkoulis. 2023. Leveraging large language mod- els for sequential recommendation. In Proceedings of the 17th ACM Conference on Recommender Systems, pages 1096-1102.
- Balázs Hidasi, Alexandros Karatzoglou, Linas Bal- trunas, and Domonkos Tikk. 2016. Session- based recommendations with recurrent neural net- works. 4th International Conference on Learning Representations, ICLR 2016.
- Ari Holtzman, Jan Buys, Li Du, Maxwell Forbes, and Yejin Choi. 2020. The curious case of neural text degeneration. 8th International Conference on Learning Representations, ICLR 2020.
- Wenyue Hua, Shuyuan Xu, Yingqiang Ge, and Yongfeng Zhang. 2023. How to index item ids for recommendation foundation models. In Proceedings of the Annual International ACM SIGIR Conference on Research and Development in Information Retrieval in the Asia Pacific Region, pages 195-204.
- Yitong Ji, Aixin Sun, Jie Zhang, and Chenliang Li. 2023. A critical study on data leakage in recommender system offline evaluation. ACM Transactions on Information Systems, 41(3):1-27.
- Wang-Cheng Kang and Julian McAuley. 2018. Self- attentive sequential recommendation. In 2018 IEEE international conference on data mining (ICDM), pages 197-206. IEEE.
- Walid Krichene and Steffen Rendle. 2020. On sampled metrics for item recommendation. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, pages 1748- 1757.
- Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gon- zalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serv- ing with pagedattention. In Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles.
- Jiayi Liao, Sihang Li, Zhengyi Yang, Jiancan Wu, Yancheng Yuan, Xiang Wang, and Xiangnan He. 2024. Llara: Large language-recommendation as- sistant. SIGIR 2024.
- Jianghao Lin et al. 2023a. How can recommender sys- tems benefit from large language models: A survey. arXiv preprint arXiv:2306.05817. · arXiv
- Xinyu Lin et al. 2023b. A multi-facet paradigm to bridge large language model and recommendation. arXiv preprint arXiv:2310.06491. · arXiv
- Ilya Loshchilov and Frank Hutter. 2019. Decou- pled weight decay regularization. 7th International Conference on Learning Representations, ICLR 2019.
- Shashank Rajput, Nikhil Mehta, Anima Singh, Raghu- nandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al. 2024. Recommender systems with generative re- trieval. Advances in Neural Information Processing Systems, 36.
- Juntao Tan, Shuyuan Xu, Wenyue Hua, Yingqiang Ge, Zelong Li, and Yongfeng Zhang. 2024. Towards llm-recsys alignment with textual id learning. SIGIR 2024.
- Jiaxi Tang and Ke Wang. 2018. Personalized top- n sequential recommendation via convolutional se- quence embedding. In Proceedings of the eleventh ACM international conference on web search and data mining, pages 565-573.
- Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems, 30.
- Ashwin K Vijayakumar, Michael Cogswell, Ram- prasath R Selvaraju, Qing Sun, Stefan Lee, David Crandall, and Dhruv Batra. 2016. Diverse beam search: Decoding diverse solutions from neural se- quence models. arXiv preprint arXiv:1610.02424. · arXiv
- Kevin Wang, Alexandre Variengien, Arthur Conmy, Buck Shlegeris, and Jacob Steinhardt. 2022. Inter- pretability in the wild: a circuit for indirect object identification in gpt-2 small. ICLR 2023.
- Yancheng Wang et al. 2023. Recmind: Large language model powered agent for recommendation. arXiv preprint arXiv:2308.14296. · arXiv
- Wei Wei, Xubin Ren, Jiabin Tang, Qinyong Wang, Lixin Su, Suqi Cheng, Junfeng Wang, Dawei Yin, and Chao Huang. 2024. Llmrec: Large language models with graph augmentation for recommendation. In WSDM 2024.
- Sean Welleck, Ilia Kulikov, Stephen Roller, Emily Dinan, Kyunghyun Cho, and Jason Weston. 2020. Neural text generation with unlikelihood train- ing. 8th International Conference on Learning Representations, ICLR 2020.
- Likang Wu et al. 2023. A survey on large lan- guage models for recommendation. arXiv preprint arXiv:2305.19860. · arXiv
- Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, et al. 2016. Google's neural machine translation system: Bridging the gap between human and machine trans- lation. arXiv preprint arXiv:1609.08144. · arXiv
- Yunjia Xi et al. 2023. Towards open-world recommen- dation with knowledge augmentation from large lan- guage models. CoRR, abs/2306.10933. · arXiv
- Zheng Yuan et al. 2023. Where to go next for rec- ommender systems? ID-vs. modality-based rec- ommender models revisited. In SIGIR 2023, pages 2639-2649. ACM.
- Jizhi Zhang, Keqin Bao, Wenjie Wang, Yang Zhang, Wentao Shi, Wanhong Xu, Fuli Feng, and Tat-Seng Chua. 2024a. Prospect personalized recommendation on large language model-based agent platform. arXiv preprint arXiv:2402.18240. · arXiv
- Junjie Zhang et al. 2023a. Recommendation as in- struction following: A large language model em- powered recommendation approach. arXiv preprint arXiv:2305.07001. · arXiv
- Yang Zhang, Keqin Bao, Ming Yan, Wenjie Wang, Fuli Feng, and Xiangnan He. 2024b. Text-like encoding of collaborative information in large language models for recommendation. ACL 2024.
- Yang Zhang, Fuli Feng, Jizhi Zhang, Keqin Bao, Qi- fan Wang, and Xiangnan He. 2023b. Collm: In- tegrating collaborative embeddings into large lan- guage models for recommendation. arXiv preprint arXiv:2310.19488. · arXiv
- Bowen Zheng et al. 2024. Adapting large language models by integrating collaborative semantics for recommendation. ICDE 2024.
Cited by
Sorted by citations
When Search Engine Services Meet Large Language Models: Visions and Challenges
Xiong, H., Bian, J., Li, Y., Li, X., Du, M., Wang, S., Yin, D., & Helal, S. (2024) · IEEE Transactions on Services Computing · doi:10.1109/tsc.2024.3451185 · Cited by 108
LLM4DSR: Leveraging Large Language Model for Denoising Sequential Recommendation
Wang, B., Liu, F., Zhang, C., Chen, J., Wu, Y., Zhou, S., Lou, X., Wang, J., Feng, Y., Chen, C., et al. (2025) · ACM Transactions on Information Systems · doi:10.1145/3762182 · Cited by 46
Self-citationReason4Rec: Deliberative User Preference Alignment of Large Language Models for Recommendation
Fang, Y. D., Wang, W. Y., Zhang, Y., Zhu, F., Wang, Q., Feng, F., & He, X. (2026) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2026.3699701 · Cited by 52
Self-citationToward Holistic Evaluation of Recommender Systems Powered by Generative Models
Deldjoo, Y., Mehta, N., Sathiamoorthy, M., Zhang, S., Castells, P., & McAuley, J. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730354 · Cited by 20
DLCRec: A Novel Approach for Managing Diversity in LLM-Based Recommender Systems
Chen, J., Gao, C., Yuan, S., Liu, S., Cai, Q., & Jiang, P. (2025) · ACM International Conference on Web Search and Data Mining (WSDM) · doi:10.1145/3701551.3703572 · Cited by 13
Consistent Explainers or Unreliable Narrators? Understanding LLM-generated Group Recommendations
Waterschoot, C., Tintarev, N., & Barile, F. (2025) · ACM Conference on Recommender Systems (RecSys) · doi:10.1145/3705328.3748015 · Cited by 7
Dual Debiasing in LLM-based Recommendation
Lu, S., Man, Z., Luo, F., & Wu, J. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3730181 · Cited by 5
Lin, X., Bao, K., Zhang, J., Zhang, Y., Wang, W., & Feng, F. (2025) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3726302.3731696 · Cited by 4
Self-citationAlgaphari, G. H., Al-Hagree, S., Al-Sanabani, M. A., Al-Shalabi, A. A., Al-Helali, B. M., & Alawdi, A. (2025) · Social Network Analysis and Mining · doi:10.1007/s13278-025-01555-3 · Cited by 3
Bringing Reasoning to Generative Recommendation Through the Lens of Cascaded Ranking
Lin, X., Liu, P., Wang, W., Hu, Y., Xu, C., Feng, F., Wang, Q., & Chua, T. (2026) · International World Wide Web Conference (WWW) · doi:10.1145/3774904.3792980 · Cited by 3
Self-citationMGFRec: Towards Reinforced Reasoning Recommendation with Multiple Groundings and Feedback
Cai, S., Gao, C., Liu, H., Shi, W., Sun, J., Tang, R., & Feng, F. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770854.3780206 · Cited by 5
Self-citationLiang, M., & Luo, Y. (2025) · ACM Conference on Conversational User Interfaces (CUI) · doi:10.1145/3719160.3737635 · Cited by 2
Beyond Tokens: A Survey on Decoding Methods for Large Language and Vision-Language Models
Wang, H., Xu, X., Yu, P. S., & Shu, K. (2026) · ACM SIGKDD Explorations Newsletter · doi:10.1145/3820356.3820357 · Cited by 3
Jiang, J., Huang, Y., Wang, Z., Wang, C., Xiong, Y., Zhang, J., & Yu, H. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817873 · Cited by 2
Self-citationTechno-bias as an AI characteristic: calibrating bias for optimal decision-making
Issa, H., & Gressieux, E. (2026) · DECISION · doi:10.1007/s40622-026-00466-2 · Cited by 1
Beyond the Single Path: Divergent Reasoning for LLM-based Recommendation
An, G., Zou, J., Yang, Y., Qin, S., Guo, W., Guo, J., & Yang, Y. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3809732 · Cited by 1
Reasoning over Semantic IDs Enhances Generative Recommendation
He, Y., Sun, Y., Tan, J., Chen, Y., Kong, X., Shen, C., Wang, X., Zhang, A., & Chua, T. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3818122 · Cited by 16
Leveraging Memory Retrieval to Enhance LLM-based Generative Recommendation
Wang, C. Y., Zhang, Y., Zhu, F., Zhang, J., Shi, T., & Feng, F. (2025) · International World Wide Web Conference (WWW) · doi:10.1145/3701716.3715596 · Cited by 1
Self-citationSelf-Monitoring Large Language Models for Click-Through Rate Prediction
Zhou, H., Yu, K., Zhang, Q., Chen, H., Zha, D., Pei, W., Kong, A., & Huang, X. X. (2025) · ACM Transactions on Information Systems · doi:10.1145/3763789 · Cited by 1
Decoding in Latent Spaces for Efficient Inference in LLM-based Recommendation
Wang, C., Zhang, Y., Wang, Z., Shi, T., Bao, K., Feng, F., & Chua, T.-S. (2025) · arXiv · doi:10.48550/arxiv.2509.11524 · Cited by 2
Self-citationLi, K., Shi, Z., Hu, M.-J., Jin, Y., Yan, X., He, F., & Wu, X. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.128726
Zheng, Y., Wang, G., Qin, J., Chen, Z., Lin, J., Wei, P., Lin, L., & Lam, K.-Y. (2025) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2025.3622687
Choi, M., Ahsan, I., Yu, H., Choe, T., & Kim, M. (2026) · Computers, materials & continua/Computers, materials & continua (Print) · doi:10.32604/cmc.2026.079504
Lu, W., Cui, X., & Yin, C. (2026) · International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) · doi:10.1145/3805712.3809758
SARE: Soft Alignment Reward for Reinforcement Learning in Generative Recommendation
Yang, Y., Wang, Z., Li, L., & Zeng, D. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3391-5_45
Hierarchical Dual Debiasing in LLM-based Recommendation
Lu, S., & Wu, J. (2026) · ACM Transactions on Information Systems · doi:10.1145/3837081
S-GRec: Personalized Semantic-Aware Generative Recommendation with Asymmetric Advantage
Jiang, J., Tang, H., Wu, W., Huang, Y., Li, Z., Li, Q., Wang, C., Zhang, J., & Yu, H. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3818473
Self-citationIntuition-Guided Latent Reasoning for LLM-Based Recommendation
Liu, C., Bai, Y., Zhao, X., Zhang, Y., Wang, Q., Feng, F., & Rong, W. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817855
Self-citationUncertainty-aware Generative Recommendation
Fan, C., Gao, C., Gong, Y., Liu, H., Feng, F., & He, X. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817975
Self-citationUnderstanding and Mitigating Semantic Popularity Bias in LLM-Based Recommender Systems
Liu, X., Tang, H., Wang, B., Cui, Y., Chen, J., Hu, B., & Wang, C. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37654-1_28
Waterschoot, C., Tintarev, N., & Barile, F. (2026) · ACM Transactions on Recommender Systems · doi:10.1145/3848640
GTP: Mitigating Popularity Bias in PLM-based Sequential Recommendation via Group-Aware Token Pruning
Yuan, R., Liu, D., Chen, H., & Ming, Z. (2026) · ACM Conference on Recommender Systems (RecSys) · doi:10.1145/3773078.3831805
Knowledge Bias in LLM-based Sequential Recommendation: Source, Persistence, and Mitigation
Dong, X., Dai, Q., Chen, N., Liu, Q., Zixuan, Y., Xie, X., Zhu, J., Dong, Z., & Wu, X.-M. (2026) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/2026.findings-emnlp.324
Co-JDPRec: Collaborative Joint Dynamic Pruning for Efficient LLM-based Generative Recommendation
Zheng, S., Jiao, P., Zhao, P., Yang, H., Zhang, D., & Wang, Y. (2026) · ACM International Conference on Information and Knowledge Management (CIKM) · doi:10.1145/3799682.3840997
Metadata
- Authors
- Bao, K., Zhang, J., Zhang, Y., Huo, X., Chen, C., & Feng, F.
- Affiliations
- University of Science and Technology of China (CN)
- National University of Singapore (SG)
- Huawei Technologies (China) (CN)
- Venue
- Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
- Published
- 2024
- Pages
- 10540-10552
- OpenAlex
- W4404782432
- Cited by
- 23
- References
- 0
- Citations by year
- 2026: 14 · 2025: 9
- FWCI
- 8.76
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- Locations
- Conference on Empirical Methods in Natural Language Processing (EMNLP) (publishedVersion, cc-by)
- Funders
- National Natural Science Foundation of China (62272437)
- University of Science and Technology of China
- Topics
- Physical Sciences › Computer Science › Information Systems › Recommender Systems and Techniques
- Physical Sciences › Computer Science › Information Systems › Information Retrieval and Search Behavior
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Keywords
- large language models, recommender systems, decoding algorithms, amplification bias, homogeneity, length normalization, recommendation diversity, item generation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
