Search-Based Interaction For Conversation Recommendation via Generative Reward Model Based Simulated User

Chunxuan Xia; Fanzhe Meng; Jinpeng Wang; Ji-Rong Wen; Junyi Li; Lei Huang; Wayne Xin Zhao; Xiaolei Wang

REVIEW 3 major objections 5 minor 45 references

Search-Based Interaction For Conversation Recommendation via Generative Reward Model Based Simulated User

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A simulated user that scores and critiques recommendations lets an untuned LLM conversational recommender beat trained baselines.

desk verdict Solid empirical integration of generative reward models into conversational recommendation, but the reward sign in Eq. 2 is ambiguous and cuts the central mechanism. read the letter →

arxiv 2504.20458 v1 pith:2XP2TEUM submitted 2025-04-29 cs.IR cs.CL

Xiaolei Wang , Chunxuan Xia , Junyi Li , Fanzhe Meng , Lei Huang , Jinpeng Wang , Wayne Xin Zhao , Ji-Rong Wen This is my paper · ORCID

classification cs.IRcs.CL

keywords conversationalrecommendationsimulatedusergenerativerewardmodelbeamsearchinstructiontuningattribute-basedcritiquingLLM-basedrecommendersystems

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

The paper sets out to show that a conversational recommender system can be improved without fine-tuning its recommender at all: the missing piece is a simulated user that behaves like a generative reward model, scoring each candidate item by the probability of an acceptance token and writing attribute-level critiques. This simulated user, called GRSU, is trained by instruction tuning on synthesized data, then used to drive multi-turn interaction with an off-the-shelf LLM recommender through beam search. The paper claims that on the ReDial and INSPIRED datasets, this interaction consistently beats all baselines, including trained conversational recommender systems, and that a general LLM paired with GRSU outperforms several fine-tuned recommender models. The reason to care is that it moves the hard part of preference understanding out of the recommender and into a feedback module that does not need ground-truth user preferences at inference time.

What carries the argument

The load-bearing object is GRSU, a single instruction-tuned language model that acts as both reward model and critic. Its two behaviors are unified by the same instruction format: generative item scoring asks it to answer 'Accept the recommendation (Yes/No)?' and the probability of the selected option token is extracted as the item's score; attribute-based item critiquing asks it to give a natural-language reaction to the recommended item list with item attributes supplied. The interaction is cast as a Markov decision process with the item list as state, the critique as action, the CRS regenerating the list as the transition, and the average generative score as the reward, and beam search over this process, followed by aggregating item scores from all visited lists, produces the final recommendation.

What would settle it

Run the same beam-search pipeline on the test conversations with the reward taken as the probability of 'Yes', the probability of 'No', and their difference; if the final rankings and Recall/NDCG values change materially, the reported gains come from an unstated sign convention rather than from learned preference, whereas stable rankings would confirm the reward direction is not the deciding factor.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that reward-guided search, powered by a simulated user, is enough to turn a general LLM into a strong conversational recommender. The simulated user GRSU is trained to provide two feedback actions: generative item scoring, where the probability of the chosen 'Yes' or 'No' token becomes the item score and the average over the list is the reward; and attribute-based item critiquing, where it says what it wants more of, such as more comedy elements. Treating an item list as a state and a critique as an action, the paper runs beam search over the interaction and then ranks all scored items across the candidate lists. GRSU reports consistent gains on ReDial and INSPIRED across Recall@k, NDCG@k, and MRR@k, with the largest gains on the low-resource INSPIRED dataset, and a simulated user trained on ReDial transfers to INSPIRED.

Load-bearing premise

The load-bearing premise is that the token probability extracted from generative item scoring is a valid reward, so that higher scores correspond to items a real user would accept; the paper does not state whether the extracted probability is the probability of 'Yes', the probability of 'No', or a signed combination, even though its illustration shows the 'No' probability used for rejected movies.

Editorial extensions

If this is right

An off-the-shelf LLM can act as a competitive conversational recommender when paired with a trained simulated user, so fine-tuning the recommender itself is not necessary.
The simulated user transfers across datasets: a user trained on the high-resource ReDial dataset improves results on the low-resource INSPIRED dataset, reducing the need for per-dataset recommender training.
The benefit appears with different LLMs of different sizes, each improving substantially when GRSU provides feedback, so the mechanism is not tied to one model.
Because one or a few interaction rounds already surpass the best trained baseline on INSPIRED, the approach offers a controllable trade-off between interaction cost and recommendation quality.

Reading between the lines

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

Beyond the paper: the same critique-plus-token-score recipe could be applied to other interactive LLM settings, such as question answering, tool use, or document drafting, where an automatic critic can guide revision under beam search.
Beyond the paper: because the evaluation uses held-out ground-truth items rather than real human satisfaction, a human study would be the next test of whether the simulated user's scores reflect actual acceptance; the paper itself does not report such a study.
Beyond the paper: resolving the sign convention of the extracted token probability is a cheap robustness check that would make the method easier to adopt, since the current write-up does not specify whether higher scores mean 'Yes' or 'No'.

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.

Desk Editor's Note

read the letter

Punchline: this is a genuinely useful integration of generative reward models into conversational recommendation, with strong and consistent empirical results, but the reward sign in Section 3.2.1 is ambiguous in a way that goes to the core mechanism, and the paper as written is not interpretable until it is fixed.

What is new: the paper's contribution is a concrete combination: an instruction-tuned simulated user that produces both generative yes/no item scores and attribute-based critiques, a beam-search interaction loop with a generic LLM-based CRS, and a cheap candidate ranking step. That combination is not just X-applied-to-Y. The experiments are extensive: four LLMs as CRSs, two datasets, ablations on both feedback behaviors, search-strategy comparisons, and a transfer study. The ablations are consistent with the design story, and the reported gains over strong baselines are large. The transfer result—a simulated user trained on ReDial and applied on INSPIRED—is a nice practical result. The 'label-free' claim is fair: no ground-truth preferences are needed at inference time; the training uses ground-truth labels from the training split, which is ordinary supervised learning, not circularity.

The soft spot is the reward definition. Eq. 2 says the item score is 'the predicted probability of the selected option token.' The prompt makes the model answer 'Accept the recommendation? Yes/No.' Figure 1(c) shows a rejected movie scored by p('No') = 0.14, and the state reward is the average of those p('No') values. Beam search then keeps states with the highest reward. If the reward is literally p('No'), search maximizes rejection. The paper never states whether the reward is p('Yes'), 1−p('No'), or a signed log-probability. The ablation says generative item scoring is the most important component, so this ambiguity is not cosmetic. Either the text and figure are mis-specified, or the search is backwards. The code link is given, but no commit hash or detailed specification appears, so a referee cannot resolve this from the manuscript. This is the kind of issue that should be fixable in revision, but it needs to be fixed.

Who this is for: people working on LLM-based conversational recommenders and user simulation. The idea of a generative reward model as a simulated user is worth taking seriously. But I would not cite it in my own work until the reward mapping is clarified and ideally the code is released in reproducible form. The paper deserves a serious referee: the problem is important, the experiments are solid, and the flaw is a specification gap rather than a fatal one. My verdict is conditional, matching the reader's.

Referee Report

3 major / 5 minor

Summary. The paper proposes GRSU, a generative reward model based simulated user for conversational recommendation systems (CRSs). The simulated user is trained via instruction tuning to perform two feedback behaviors: generative item scoring (outputting 'Accept the recommendation? Yes/No' with token probabilities) and attribute-based item critique (producing natural-language critiques of recommended items). The interaction between a zero-shot LLM-based CRS and the simulated user is cast as a Markov decision process and searched with beam search; a final efficient ranking step aggregates item scores across the searched candidate lists. Experiments on ReDial and INSPIRED report consistent improvements over trained CRS baselines and multiple zero-shot LLMs, with ablations showing each component contributes, and a transferability study showing a simulated user trained on high-resource ReDial helps on low-resource INSPIRED.

Significance. If the mechanism is sound, the paper makes a useful contribution: it shows that a fine-tuned simulated user can provide label-free feedback at inference time to guide zero-shot LLM CRSs, achieving strong gains over trained CRSs on two benchmarks. The experimental design is extensive: main results across six metrics, ablations over components and implementation variants, transferability analysis, robustness across four LLM backbones, and a comparison of search strategies. The code link is provided, which aids reproducibility. The central weakness is the ambiguity of the scalar reward extracted from the generative item scoring, which touches the core search-and-ranking mechanism; because the ablation in Section 4.3.1 identifies this component as the most important, the ambiguity must be resolved before the empirical claims can be fully interpreted.

major comments (3)

[Section 3.2.1, Eq. (2), Figure 1(c), Section 3.3] The definition of the generative item score is ambiguous and potentially inverted. The text states that 'the predicted probability of the selected option token is then interpreted as the item score,' and Figure 1(c) explicitly labels the scores of rejected items as p('No') = 0.14 and 0.08, with an average of 0.11. However, Section 3.3 uses this score as the reward in beam search, retaining states with the highest rewards, and the candidate ranking method sorts items by the same scores. If the selected token is 'No', then a high p('No') corresponds to confident rejection, so maximizing the reward would select the most strongly rejected items and rank them at the top. The paper never states whether the implemented score is p('Yes'), p('No'), 1 - p('No'), or a signed transform, nor does it reconcile the figure with the text. Since the ablation in Section 4.3.1 shows that generative item scoring is the single most important component, this ambiguity is load-bearing for the central claim. Please specify the exact post-processor Phi_s, correct the text and Figure 1(c) accordingly, and confirm which quantity is maximized in beam search and sorted in the final ranking.
[Section 4.1 and Section 4.3.3, Tables 3 and 5] The main results in Table 3 do not state which data are used to train the simulated user for each evaluation dataset. For INSPIRED, the main result (Recall@10 = 0.352) exactly matches the 'ReDial+INSPIRED' row of Table 5, indicating that the simulated user is trained on both datasets for that experiment. For ReDial, it is unclear whether the simulated user is trained on ReDial only or also on INSPIRED. Please state explicitly the training data configuration used for each row of Table 3, and discuss how the transferability claim is affected if the low-resource INSPIRED evaluation already includes training on the high-resource ReDial data (which is legitimate for transfer but should be transparent).
[Section 3.3] The same simulated user generates both the critique actions (Eq. 3) and the reward scores (Eq. 2). Because the reward and the action come from the same model, there is a risk that the reward rewards states that the model itself steers toward, rather than states a real user would accept. This potential self-bias is not discussed. Please provide an analysis or argument showing that the reward is a faithful proxy for real user acceptance despite being produced by the same model that generates the critiques, for example by comparing reward-based rankings with held-out ground-truth acceptance.

minor comments (5)

[Table 3 and Section 4.1] The baseline name is misspelled as 'Populalarity' in both the table and the baselines list; it should be 'Popularity'.
[Section 3.2.2] The text says 'we sample k negative items from the set I - I_k' but the implementation section says 'we sample one additional item as a negative sample for each ground-truth item.' The symbol k is already used for the turn index; please clarify the number of negative samples and avoid reusing k.
[Figure 1(c)] The caption states that the probability of the reward token (i.e., 'No') is extracted as the score. If the intended score is p('Yes') or a transformed quantity, the caption should be corrected; otherwise, the example should be reconciled with the claimed reward objective.
[Section 4.1] The phrase 'bf16 16-bit mixed precision training' is redundant; consider rewriting as 'bf16 mixed precision training'.
[Implementation details] The code link is given, but no commit hash or direct pointer to the reward extraction function is provided. Adding a commit hash and a reference to the Phi_s implementation would resolve the ambiguity in Eq. (2) for readers and reviewers.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the simulated user is trained on ground-truth labels but evaluated on held-out items, and the reward-guided search is not equivalent to its inputs by construction.

full rationale

The paper's derivation chain is self-contained rather than circular. The simulated user is trained via instruction tuning on data synthesized from ground-truth items and user preferences (Section 3.2.2), but during evaluation the model receives only conversation history and recommended item lists; ground-truth preferences are not available at inference time, as stated in Section 3.2.2 and Section 3.3. The final recommendation quality is measured against ground-truth items in the held-out test sets (Section 4.2), not against the simulated user's own scores. The generative item scoring reward (Eq. 2) is a fitted model trained on acceptance labels, and using that fitted score to guide beam search and candidate ranking is a legitimate modeling choice, not a prediction forced by construction. The paper's self-citations, such as prior work on CRS evaluation, are not load-bearing for the central claim; no uniqueness theorem or prior result is invoked to forbid alternative designs. One concern noted in the review is that Figure 1(c) shows the extracted score for rejected items as the 'No' token probability, while the reward-guided search retains states with highest rewards, which would be nonsensical if the score were literally p('No'). This is an implementation/interpretation ambiguity and a correctness risk, not a circularity: the paper's equations do not reduce to their inputs by definition. Therefore, no circularity step meets the evidence bar for flagging.

Assumptions & free parameters 5 free parameters · 3 assumptions · 1 invented entities

The central claim rests on the simulated user's reward being a valid proxy for acceptance. The search hyperparameters are chosen by hand rather than derived, and the training data for critiques are generated by an LLM using ground-truth preferences. The paper does not provide independent evidence for the simulated user's fidelity to real users.

free parameters (5)

beam_width B = 4
Chosen by the authors for the search process; it affects how many candidate item lists survive each step.
expansion_width N = 4
Set equal to beam width; it controls the number of sampled actions per state in the beam search.
search_depth D = 5
The number of interaction rounds in the beam search; performance grows with D, so the choice shapes the reported gains.
item_list_length = 10
All item lists are fixed to length 10, and evaluation uses Recall@10 and NDCG@10, so this choice directly shapes results.
number_of_negative_samples_per_ground_truth_item = 1
One negative item is sampled per positive item during data synthesis for generative item scoring; this affects the training signal.

assumptions (3)

domain assumption The ground-truth items in ReDial and INSPIRED are a valid proxy for real user acceptance of recommendations.
The method and evaluation both define success as matching dataset ground-truth items, so the simulated user's feedback is optimized against this proxy rather than against measured user satisfaction. This enters at Section 4.1.
domain assumption The LLM-generated critiques used as training labels for attribute-based item critiquing are representative of real user feedback.
Instruction data for critiques is synthesized by Llama-3.3-70B using ground-truth preferences, not collected from real users; transfer to real user behavior is assumed. This is stated in Section 3.2.2.
ad hoc to paper The probability of the selected option token is a valid scalar reward for search.
The paper does not define a mapping from p(Yes) and p(No) to a reward, and Figure 1(c) suggests p(No) is used, which would invert acceptance. This assumption is load-bearing in Sections 3.2.1 and 3.3.

invented entities (1)

GRSU (Generative Reward model based Simulated User)
purpose: A trained LLM agent that provides item scores and attribute critiques to CRSs during multi-turn interaction.
The paper introduces this agent and evaluates it only on the same datasets used for training; it provides no external validation against real user behavior or independent tasks.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Search-Based Interaction For Conversation Recommendation via Generative Reward Model Based Simulated User." pith.science (2026). https://pith.science/paper/2XP2TEUM

@misc{pith2026250420458,
  author       = {Pith},
  title        = {Pith review of: Search-Based Interaction For Conversation Recommendation via Generative Reward Model Based Simulated User},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2XP2TEUM}},
  note         = {Machine review of arXiv:2504.20458}
}

read the original abstract

Conversational recommendation systems (CRSs) use multi-turn interaction to capture user preferences and provide personalized recommendations. A fundamental challenge in CRSs lies in effectively understanding user preferences from conversations. User preferences can be multifaceted and complex, posing significant challenges for accurate recommendations even with access to abundant external knowledge. While interaction with users can clarify their true preferences, frequent user involvement can lead to a degraded user experience. To address this problem, we propose a generative reward model based simulated user, named GRSU, for automatic interaction with CRSs. The simulated user provides feedback to the items recommended by CRSs, enabling them to better capture intricate user preferences through multi-turn interaction. Inspired by generative reward models, we design two types of feedback actions for the simulated user: i.e., generative item scoring, which offers coarse-grained feedback, and attribute-based item critique, which provides fine-grained feedback. To ensure seamless integration, these feedback actions are unified into an instruction-based format, allowing the development of a unified simulated user via instruction tuning on synthesized data. With this simulated user, automatic multi-turn interaction with CRSs can be effectively conducted. Furthermore, to strike a balance between effectiveness and efficiency, we draw inspiration from the paradigm of reward-guided search in complex reasoning tasks and employ beam search for the interaction process. On top of this, we propose an efficient candidate ranking method to improve the recommendation results derived from interaction. Extensive experiments on public datasets demonstrate the effectiveness, efficiency, and transferability of our approach.

Figures

Figures reproduced from arXiv: 2504.20458 by the authors.

**Figure 2.** Ablation study on the INSPIRED dataset. “critiquing” refers to attribute-based item critiquing, “scoring” refers to generative item scoring, and “ranking” refers to efficient candidate ranking. utilize other datasets (e.g., ReDial) for training, effectively mitigating the challenge of data scarcity. For this point, we provide a detailed analysis in Section 4.3.3. Furthermore, our approach employs a general LLM as t… view at source ↗

**Figure 3.** The performance changes with respect to the depth [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 20 canonical work pages

[1]

Marah Abdin, Jyoti Aneja, Harkirat Behl, Sébastien Bubeck, Ronen Eldan, Suriya Gunasekar, Michael Harrison, Russell J Hewett, Mojan Javaheripi, Piero Kauff- mann, et al. 2024. Phi-4 technical report. arXiv preprint arXiv:2412.08905 (2024)

arXiv 2024
[2]

Wanling Cai and Li Chen. 2019. Towards a Taxonomy of User Feedback Intents for Conversational Recommendations. RecSys (Late-Breaking Results) 2431 (2019), 51–55

work page 2019
[3]

Wanling Cai and Li Chen. 2020. Predicting user intents and satisfaction with dialogue-based conversational recommendations. In Proceedings of the 28th ACM Conference on User Modeling, Adaptation and Personalization . 33–42

work page 2020
[4]

Yupeng Chang, Xu Wang, Jindong Wang, Yuan Wu, Linyi Yang, Kaijie Zhu, Hao Chen, Xiaoyuan Yi, Cunxiang Wang, Yidong Wang, et al . 2024. A survey on evaluation of large language models. ACM Transactions on Intelligent Systems and Technology 15, 3 (2024), 1–45

2024
[5]

Li Chen and Pearl Pu. 2012. Critiquing-based recommenders: survey and emerg- ing trends. User Modeling and User-Adapted Interaction 22 (2012), 125–150

work page 2012
[6]

Tianqi Chen, Bing Xu, Chiyuan Zhang, and Carlos Guestrin. 2016. Training deep nets with sublinear memory cost. arXiv preprint arXiv:1604.06174 (2016)

arXiv 2016
[7]

Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168 (2021)

arXiv 2021
[8]

Huy Dao, Yang Deng, Dung D Le, and Lizi Liao. 2024. Broadening the view: Demonstration-augmented prompt learning for conversational recommendation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval . 785–795

work page 2024

Show all 45 references

[9]

Tri Dao. [n. d.]. FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning. In The Twelfth International Conference on Learning Represen- tations
[10]

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 (2024)

2024 arXiv
[11]

Chongming Gao, Wenqiang Lei, Xiangnan He, Maarten de Rijke, and Tat-Seng Chua. 2021. Advances and challenges in conversational recommender systems: A survey. AI open 2 (2021), 100–126

2021
[12]

Shirley Anugrah Hayati, Dongyeop Kang, Qingxiaoyang Zhu, Weiyan Shi, and Zhou Yu. 2020. Inspired: Toward sociable recommendation dialog systems. arXiv preprint arXiv:2009.14306 (2020)

2020 arXiv
[13]

Zhankui He, Zhouhang Xie, Rahul Jha, Harald Steck, Dawen Liang, Yesu Feng, Bodhisattwa Prasad Majumder, Nathan Kallus, and Julian McAuley. 2023. Large language models as zero-shot conversational recommenders. InProceedings of the 32nd ACM international conference on informatio...

2023
[14]

Zhankui He, Zhouhang Xie, Harald Steck, Dawen Liang, Rahul Jha, Nathan Kallus, and Julian McAuley. 2024. Reindex-Then-Adapt: Improving Large Language Models for Conversational Recommendation. arXiv preprint arXiv:2405.12119 (2024)

2024 arXiv
[15]

Chen Huang, Peixin Qin, Yang Deng, Wenqiang Lei, Jiancheng Lv, and Tat-Seng Chua. 2024. Concept–An Evaluation Protocol on Conversation Recommender Systems with System-and User-centric Factors. arXiv preprint arXiv:2404.03304 (2024)

2024 arXiv
[16]

Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, et al. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276 (2024)

2024 arXiv
[17]

Dietmar Jannach, Ahtsham Manzoor, Wanling Cai, and Li Chen. 2021. A survey on conversational recommender systems. ACM Computing Surveys (CSUR) 54, 5 (2021), 1–36

2021
[18]

Yucheng Jin, Li Chen, Wanling Cai, and Pearl Pu. 2021. Key qualities of conver- sational recommender systems: From users’ perspective. In Proceedings of the 9th International Conference on Human-Agent Interaction . 93–102

2021
[19]

Yoon Kim. 2014. Convolutional Neural Networks for Sentence Classification. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing, EMNLP 2014, October 25-29, 2014, Doha, Qatar, A meeting of SIGDAT, a Special Interest Group of the ACL , Alessa...

2014 doi
[20]

Wenqiang Lei, Xiangnan He, Yisong Miao, Qingyun Wu, Richang Hong, Min- Yen Kan, and Tat-Seng Chua. 2020. Estimation-action-reflection: Towards deep interaction between conversational and recommender systems. In Proceedings of the 13th International Conference on Web Search and...

2020
[21]

Wenqiang Lei, Gangyi Zhang, Xiangnan He, Yisong Miao, Xiang Wang, Liang Chen, and Tat-Seng Chua. 2020. Interactive path reasoning on graph for conver- sational recommendation. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining ....

2020
[22]

Raymond Li, Samira Ebrahimi Kahou, Hannes Schulz, Vincent Michalski, Laurent Charlin, and Chris Pal. 2018. Towards deep conversational recommendations. Advances in neural information processing systems 31 (2018)

2018
[23]

Shijun Li, Wenqiang Lei, Qingyun Wu, Xiangnan He, Peng Jiang, and Tat-Seng Chua. 2021. Seamlessly unifying attributes and items: Conversational recommen- dation for cold-start users. ACM Transactions on Information Systems (TOIS) 39, 4 (2021), 1–29

2021
[24]

Wendi Li, Wei Wei, Xiaoye Qu, Xian-Ling Mao, Ye Yuan, Wenfeng Xie, and Dan- gyang Chen. 2023. TREA: Tree-Structure Reasoning Schema for Conversational Recommendation. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Pap...

2023
[25]

Hunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. [n. d.]. Let’s Verify Step by Step. In The Twelfth International Conference on Learning Representations
[26]

Dakota Mahan, Duy Van Phung, Rafael Rafailov, Chase Blagden, Nathan Lile, Louis Castricato, Jan-Philipp Fränken, Chelsea Finn, and Alon Albalak. 2024. Generative reward models. arXiv preprint arXiv:2410.12832 (2024)

2024 arXiv
[27]

Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra Search-Based Interaction For Conversation Recommendation via Generative Reward Model Based Simulated User SIGIR ’25, Jul...

2023
[28]

Jeff Rasley, Samyam Rajbhandari, Olatunji Ruwase, and Yuxiong He. 2020. Deep- speed: System optimizations enable training deep learning models with over 100 billion parameters. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining ...

2020
[29]

Charlie Snell, Jaehoon Lee, Kelvin Xu, and Aviral Kumar. 2024. Scaling llm test- time compute optimally can be more effective than scaling model parameters. arXiv preprint arXiv:2408.03314 (2024)

2024 arXiv
[30]

Ziyu Wan, Xidong Feng, Muning Wen, Stephen Marcus McAleer, Ying Wen, Weinan Zhang, and Jun Wang. 2024. Alphazero-like tree-search can guide large language model decoding and training. In Forty-first International Conference on Machine Learning

2024
[31]

Xiaolei Wang, Xinyu Tang, Wayne Xin Zhao, Jingyuan Wang, and Ji-Rong Wen
[32]

Xiaolei Wang, Kun Zhou, Xinyu Tang, Wayne Xin Zhao, Fan Pan, Zhao Cao, and Ji-Rong Wen. 2023. Improving conversational recommendation systems via counterfactual data simulation. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . 2398–2408

2023
[33]

Xiaolei Wang, Kun Zhou, Ji-Rong Wen, and Wayne Xin Zhao. 2022. Towards unified conversational recommender systems via knowledge-enhanced prompt learning. In Proceedings of the 28th ACM SIGKDD conference on knowledge discov- ery and data mining . 1929–1937

2022
[34]

Zhouhang Xie, Junda Wu, Hyunsik Jeon, Zhankui He, Harald Steck, Rahul Jha, Dawen Liang, Nathan Kallus, and Julian McAuley. 2024. Neighborhood-Based Collaborative Filtering for Conversational Recommendation. In Proceedings of the 18th ACM Conference on Recommender Systems . 1045–1050

2024
[35]

Bowen Yang, Cong Han, Yu Li, Lei Zuo, and Zhou Yu. 2022. Improving Con- versational Recommendation Systems’ Quality with Context-Aware Item Meta- Information. In Findings of the Association for Computational Linguistics: NAACL

2022
[36]

Ting Yang and Li Chen. 2024. Unleashing the Retrieval Potential of Large Lan- guage Models in Conversational Recommender Systems. In Proceedings of the 18th ACM Conference on Recommender Systems . 43–52

2024
[37]

Se-eun Yoon, Zhankui He, Jessica Echterhoff, and Julian McAuley. 2024. Evaluat- ing Large Language Models as Generative User Simulators for Conversational Recommendation. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational L...

2024
[38]

Lunjun Zhang, Arian Hosseini, Hritik Bansal, Mehran Kazemi, Aviral Kumar, and Rishabh Agarwal. [n. d.]. Generative Verifiers: Reward Modeling as Next-Token Prediction. In The 4th Workshop on Mathematical Reasoning and AI at NeurIPS’24
[39]

Shengyu Zhang, Linfeng Dong, Xiaoya Li, Sen Zhang, Xiaofei Sun, Shuhe Wang, Jiwei Li, Runyi Hu, Tianwei Zhang, Fei Wu, et al. 2023. Instruction tuning for large language models: A survey. arXiv preprint arXiv:2308.10792 (2023)

2023
[40]

Xiaoyu Zhang, Xin Xin, Dongdong Li, Wenxuan Liu, Pengjie Ren, Zhumin Chen, Jun Ma, and Zhaochun Ren. 2023. Variational reasoning over incomplete knowl- edge graphs for conversational recommendation. In Proceedings of the Sixteenth ACM International Conference on Web Search and...

2023
[41]

Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, et al. 2023. A Survey of Large Language Models. arXiv preprint arXiv:2303.18223 (2023)

2023 arXiv
[42]

Zhipeng Zhao, Kun Zhou, Xiaolei Wang, Wayne Xin Zhao, Fan Pan, Zhao Cao, and Ji-Rong Wen. 2023. Alleviating the long-tail problem in conversational recommender systems. InProceedings of the 17th ACM Conference on Recommender Systems. 374–385

2023
[43]

Kun Zhou, Xiaolei Wang, Yuanhang Zhou, Chenzhan Shang, Yuan Cheng, Wayne Xin Zhao, Yaliang Li, and Ji-Rong Wen. 2021. CRSLab: An Open-Source Toolkit for Building Conversational Recommender System. In Proceedings of the 59th Annual Meeting of the Association for Computational L...

2021
[44]

Lixi Zhu, Xiaowen Huang, and Jitao Sang. 2024. A LLM-based Controllable, Scalable, Human-Involved User Simulator Framework for Conversational Rec- ommender Systems. arXiv preprint arXiv:2405.08035 (2024)

2024 arXiv
[2023]

In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing

Rethinking the Evaluation for Conversational Recommendation in the Era of Large Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing . 10052–10065

2023

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Marah Abdin, Jyoti Aneja, Harkirat Behl, Sébastien Bubeck, Ronen Eldan, Suriya Gunasekar, Michael Harrison, Russell J Hewett, Mojan Javaheripi, Piero Kauff- mann, et al. 2024. Phi-4 technical report. arXiv preprint arXiv:2412.08905 (2024)

arXiv 2024

[2] [2]

Wanling Cai and Li Chen. 2019. Towards a Taxonomy of User Feedback Intents for Conversational Recommendations. RecSys (Late-Breaking Results) 2431 (2019), 51–55

work page 2019

[3] [3]

Wanling Cai and Li Chen. 2020. Predicting user intents and satisfaction with dialogue-based conversational recommendations. In Proceedings of the 28th ACM Conference on User Modeling, Adaptation and Personalization . 33–42

work page 2020

[4] [4]

Yupeng Chang, Xu Wang, Jindong Wang, Yuan Wu, Linyi Yang, Kaijie Zhu, Hao Chen, Xiaoyuan Yi, Cunxiang Wang, Yidong Wang, et al . 2024. A survey on evaluation of large language models. ACM Transactions on Intelligent Systems and Technology 15, 3 (2024), 1–45

2024

[5] [5]

Li Chen and Pearl Pu. 2012. Critiquing-based recommenders: survey and emerg- ing trends. User Modeling and User-Adapted Interaction 22 (2012), 125–150

work page 2012

[6] [6]

Tianqi Chen, Bing Xu, Chiyuan Zhang, and Carlos Guestrin. 2016. Training deep nets with sublinear memory cost. arXiv preprint arXiv:1604.06174 (2016)

arXiv 2016

[7] [7]

Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168 (2021)

arXiv 2021

[8] [8]

Huy Dao, Yang Deng, Dung D Le, and Lizi Liao. 2024. Broadening the view: Demonstration-augmented prompt learning for conversational recommendation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval . 785–795

work page 2024