Pith. sign in

REVIEW 4 major objections 5 minor 4 cited by

Are Longer Prompts Always Better? Prompt Selection in Large Language Models for Recommendation Systems

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that no single prompt design dominates for LLM-based recommendation and that selecting a prompt per dataset with a small validation set, guided by a relative-performance analysis, beats fixed baselines and full grid…

desk verdict A useful large empirical grid for LLM prompt selection, but the selection method's advantage over simple grid search is not yet demonstrated because the validation protocol is ambiguous and there are no error bars. read the letter →

arxiv 2412.14454 v1 pith:PFEH4MAD submitted 2024-12-19 cs.IR cs.CL

Genki Kusano , Kosuke Akimoto , Kunihiro Takeoka This is my paper · ORCID
classification cs.IRcs.CL
keywords promptselectionlargelanguagemodelsrecommendationsystemsrelativeperformanceindicatorzero-shotrankingvalidationdatanDCG@10cost-efficientinference
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 asks whether longer or more elaborate prompts are always better in LLM-based recommendation systems, and answers no. Across 450 experiments with 90 prompt variants and five datasets, it finds that the best prompt changes with the dataset, with format, sampling size, and item attributes all interacting. It proposes selecting prompts with a small validation set: a relative-performance analysis narrows the field to two candidate prompts, and the better one on 100 validation users is used for test inference. This selection ranks first in nDCG@10 on three of five datasets and second on one, beating four previous prompt designs and even full grid search. A cost-aware variant, which searches with a cheap LLM and rescoring where needed, is highest or tied on four of five datasets.

What carries the argument

The load-bearing mechanism is the relative performance indicator (RPI). For each prompt component (sampling format, sample size, item attributes), the paper fixes the other components, computes the ratio of nDCG@10 when the component changes, and averages these ratios across formats; RPI is that average minus one, expressed as a percentage, with positive values meaning the component helps. RPI turns a 90-variant prompt grid into ranked component preferences and lets the authors construct two candidate prompts, one for sampling-based and one for summarization-based prompts. The second mechanism is validation-set prompt selection: the candidate that scores higher on 100 held-out users is chosen, and the paper's cost-aware variant uses a cheaper LLM for exploration and a stronger LLM to rescore candidates on validation before final inference.

What would settle it

Read the code or protocol to determine which 100 users produced the RPI rankings; if those rankings used the test users, the reported gains are in-sample rather than forward-looking. A second check is to rerun the selection on ten fresh 100-user validation splits: if the chosen prompt and the resulting test nDCG@10 change substantially across splits, the single-split protocol does not support the claimed advantage.

Watch

Extended reading notes

Core claim

The central discovery is that prompt effectiveness in LLM recommendation is conditional, not intrinsic: no component consistently wins, and accuracy differences are large, with the minimum near random in some datasets. The paper's constructive result is an automatic prompt-selection procedure. It first builds a grid of 90 prompts by crossing item-attribute sets (title, category, description and their combinations) with sampling schemes (Random, Latest, Extract), sample sizes $k = 5, 10, 20, 30$, and summarization. On validation data it computes a relative performance indicator (RPI) for each component, constructing one prompt from the best sampling-side combination and one from the best summarization-side combination; the better of the two on validation is then used for test inference. In the paper's experiments this RPI-guided pick outperformed the four prior prompt designs in most datasets and, notably, outperformed selecting the single best prompt on validation (grid search), which the paper attributes to overfitting the validation set. It also shows that a stronger LLM can be reserved for final inference while a cheaper LLM does the search, because rescoring the few candidate prompts on validation with the stronger model yields the best or tied-best accuracy in four of five datasets.

Load-bearing premise

The reported gains depend on a protocol detail the paper never states clearly: the relative-performance analysis that picks the two candidate prompts must be run on the 100 validation users, not on the 100 test users, and one 100-user split must be enough to rank 90 prompt designs reliably.

Editorial extensions

If this is right

  • A practitioner can select a competitive prompt for a new dataset by running about 100 validation users instead of testing all 90 prompts on the full test set.
  • Title-only prompts, the focus of much earlier work, are not universally best; adding categories or descriptions improves some datasets, while categories alone consistently hurt.
  • Full grid search on validation data can overfit; a component-based aggregation such as RPI can beat it on test data.
  • Exploration cost can be cut by doing the search with a cheaper LLM and the final inference with a stronger one; the paper's cost-aware strategy is best or tied in four of five datasets.
  • Because no prompt transfers across datasets or LLMs, prompt choice should be treated as a per-dataset, per-model decision rather than a fixed template.

Reading between the lines

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

  • The RPI factorization is not specific to recommendation: any prompt family with independent components (format, size, content) could be ranked the same way, so the method may transfer to other LLM tasks.
  • The paper does not test repeated validation splits; a natural next experiment is to resample the 100-user validation set many times and check whether the selected prompt and test nDCG@10 are stable, since large variance would undermine the single-split protocol.
  • The dataset statistics reported in the paper (title length, category informativeness, description availability) correlate with which components help, suggesting a future 'prompt predictor' that skips validation entirely by mapping dataset statistics to prompt components.
  • The pruning suggestion in the discussion implies an even cheaper protocol: run 36 prompts instead of 90 on the cheap LLM, then rescore survivors with the strong LLM; this is not evaluated in the paper.
Share X Bluesky LinkedIn Reddit HN

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. The paper asks whether prompt design choices for LLM-based recommender systems can be selected systematically. It standardizes an inference prompt template, varies the user-information component along three dimensions (sampling format Random/Latest/Extract, sampling size k, item attributes T/C/D combinations, and summarization), and evaluates 90 prompts on 100-user test sets from five Amazon categories using gpt-4o-mini. The authors introduce the Relative Performance Indicator (RPI), an average accuracy-ratio summary over the prompt grid, to identify which component choices help. They then propose selecting prompts with a separate 100-user validation set, either by grid search (GS) or by constructing two RPI-guided candidate prompts and choosing between them (RPI). They report that RPI-guided selection ranks first in three datasets and second in one (Table 5), and that a gpt-4o-based cost-aware variant (GS*) is best or tied for best in four of five datasets (Table 6). The paper also reports API costs and discusses limitations around model generality and exploration cost.

Significance. If the findings are stable, the paper provides a useful empirical map of a largely unexplored design space and a low-cost validation procedure for prompt selection. The strengths are the systematic 90-prompt grid, the standardized template that isolates the user-information component, the explicit accounting of API calls and cost, and a clear statement of limitations (Section 5). The central claims, however, rest on two things the manuscript does not yet establish: a clean separation between the data used to derive the RPI-guided candidates and the data used to evaluate them, and evidence that the reported differences are not noise from a single 100-user split. Neither issue is a conceptual showstopper, and both can be fixed with additional reporting or experiments, but they are load-bearing for the headline 'validation set is enough' claim.

major comments (4)
  1. [§4.2–4.3, Tables 3–5] The RPI-guided selection protocol is not fully specified. Section 4.2 reports RPI tables and component-level conclusions computed on the 100 users selected in Section 4.1; Section 4.1 does not state that these users are a test set, but Table 5 later evaluates selected prompts on exactly these users. Section 4.3 introduces 100 different validation users and says 'we conducted an RPI analysis ... created two prompts with the best combinations,' but it never states explicitly that this second RPI analysis was recomputed on the validation users only. If the candidate prompts in Table 5 were selected using RPI rankings from the test users, then the validation step only chooses between two prompts already known to be good on test data, and the RPI row in Table 5 is an in-sample comparison. The same ambiguity affects the cost-reduction proposal in Section 5, which suggests excluding randomly sampled, k=5, C, CD, and Extract prompts on the basis of RPI patterns from Section 4.2. Please state, for each RPI table, the exact user set used, and ideally release the split definitions; the core claim that a 100-user validation set suffices depends on this.
  2. [Tables 2–6, §4.3] All reported nDCG@10 values are point estimates from a single 100-user split, with no variance, confidence intervals, significance tests, or repeated seeds for the random negative sampling described in Section 3.1. Differences between top methods are often small, e.g., Table 5 Grocery RPI and GS both 0.438, Table 5 Movie RPI 0.624 vs. L-10-T 0.646, and Table 6 Music RPI 0.723 vs. GS* 0.723. Without an estimate of split-to-split variability, the claims that RPI 'ranked first in three out of five cases' (Section 4.3) and that GS* 'outperformed previous research in four out of five datasets' (Section 4.4) are not yet supported. Please add repeated splits, bootstrap confidence intervals, or paired significance tests for the main comparisons.
  3. [§4.2, Figure 3, Tables 3–4] The RPI is defined as an average of accuracy ratios over the grid (Figure 3) and is reported as a single number. The text interprets positive values as indicating that a component is 'statistically superior' (Section 4.2), but no variance estimate or test is provided; moreover, the 72 or 18 grid entries are not independent because they share prompts, datasets, and users. This interpretation matters because the RPI-guided candidate construction in Section 4.3 relies on these point estimates to pick components (e.g., L-20-TCD, SR-30-TCD). Please either add uncertainty estimates or soften the statistical language to 'associated with higher average accuracy in this grid.'
  4. [§4.4, Table 6] The GS* procedure is not fully described. It is said to select the prompt with highest accuracy on validation data using gpt-4o among the four previous prompts, GS, and RPI; however, the validation accuracy numbers and the exact candidate set are not reported, and no repeated selection is shown. Since Table 6 is the basis for the cost-efficiency claim, please report the validation-set accuracies and show that the gpt-4o selection is stable across splits or at least report the variability of the chosen prompt.
minor comments (5)
  1. [Page 2, author line] 'Kusano at al.' should be 'Kusano et al.'
  2. [Tables 1–6] Several table captions and headings contain formatting artifacts such as 'T able' and 'T able 1'; these should be corrected.
  3. [Section 4.1] The process for generating the 18 negative items per user is described as random, but the seeds and the relationship between the negative sampling and the evaluation split are not reported; please document this for reproducibility.
  4. [Section 3.2 and Table 4] The abbreviations SR, SL, and SE are used in Table 4 before their component notation is fully explained in the text; a one-line definition in Section 3.2 would help.
  5. [Section 3.1] The phrase 'the latest two items' is used for positive examples, while Figure 1 shows ten viewed items; the relationship between the number of history items, the sampling size k, and the candidate list should be clarified.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the prompt-selection pipeline uses disjoint validation and test users with external baselines, and the RPI analysis is a diagnostic applied on validation data rather than a fitted target renamed as a prediction.

full rationale

Walking the paper's derivation chain, the central evaluation is an out-of-sample comparison: Section 4.1 defines 100 users, Section 4.3 explicitly introduces '100 users different from those described in Section 4.1' as validation users, and Tables 5–6 report nDCG@10 on the test users for prompts selected via validation-based grid search (GS), RPI-guided selection, and external prior-work prompts. RPI is an interpretable main-effect summary of the 90-prompt accuracy grid, not a target variable; it is used to construct two candidate prompts on validation data, after which the better candidate is selected on validation and reported on test. No equation in the paper defines the reported test nDCG@10 as a function of the validation scores or RPI values. The only nearby concern is a protocol ambiguity about which users produced the Section 4.2 RPI tables; however, the text introduces validation users before describing the RPI-based candidate construction in Section 4.3, and the RPI prompt names in Table 5 (e.g., Music: SR-30-TCD) differ from the best-RPI prompts in Tables 3 and 4 (L-20-TCD and SL-30-TCD), indicating the RPI analysis was rerun on the validation data. That ambiguity is a transparency and reproducibility matter, not a definitional circularity. The paper cites only external prior work, so no self-citation chain is load-bearing. The GS* row in Table 6 is a selection rule—choosing the candidate with highest validation accuracy—and its test values are measured on held-out test users, so it is not a fitted input renamed as a prediction. Because no circular reduction can be exhibited from the paper's own equations or citations, the circularity score is 0.

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

The central claim rests on a hand-built grid of 90 prompt variants, a single 100-user validation split per dataset, no published code or exact prompt templates, and the RPI statistic without variance estimates. No model parameters are fitted; the listed free parameters are experimental design choices that the conclusions depend on.

free parameters (5)
  • Validation set size = 100 users per dataset
    Chosen by hand; no ablation on size, yet the abstract claims minimal validation data.
  • Positive/negative candidate ratio = 2 positive / 18 negative per user
    Defines the task and the random baseline (nDCG around 0.27); no sensitivity analysis is provided.
  • LLM sampling temperature = 0.3
    Decoding temperature for gpt-4o-mini; its effect on ranking stability is not analyzed.
  • Sampling grid sizes = k in {5,10,20,30}; summarization k=30 or 100
    Defines the 90-prompt search space, so conclusions apply only within this grid.
  • Evaluation cutoff = nDCG@10
    Choice of ranking horizon; no sensitivity check is reported.
assumptions (5)
  • domain assumption The standardized prompt template with role prefix, task conditions, and repeated instructions is a neutral base that does not favor any user-information variant.
    Introduced in Section 3.1; all comparisons assume template effects are additive and independent of the user-information component.
  • domain assumption Users with 30+ ratings and positive items rated 3+ stars represent a meaningful LLM-RS evaluation population.
    Section 4.1 filters out cold-start users, which is one of the motivating applications, so the evaluation does not actually test cold-start performance.
  • ad hoc to paper RPI, as an average of accuracy ratios over the prompt grid, is a meaningful indicator of component superiority without variance estimates.
    Defined in Section 4.2; the paper introduces RPI but provides no confidence intervals or significance tests.
  • domain assumption Prompt rankings on 100 validation users transfer to the 100 test users.
    Section 4.3 relies on one small split; no repeated splits, bootstrap, or error bars support this transfer assumption.
  • domain assumption Findings with gpt-4o-mini and gpt-4o generalize to other LLMs.
    Stated as an unverified limitation in Section 5, yet the paper frames its guidelines broadly.
invented entities (1)
  • Relative Performance Indicator (RPI)
    purpose: Summarize whether a prompt component (sampling format, size, item attribute) helps or hurts by averaging accuracy ratios across the grid.
    Paper-specific statistic with no external benchmark, no calibration, and no predictive validation outside the same experiment grid.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Are Longer Prompts Always Better? Prompt Selection in Large Language Models for Recommendation Systems." pith.science (2026). https://pith.science/paper/PFEH4MAD

@misc{pith2026241214454,
  author       = {Pith},
  title        = {Pith review of: Are Longer Prompts Always Better? Prompt Selection in Large Language Models for Recommendation Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PFEH4MAD}},
  note         = {Machine review of arXiv:2412.14454}
}
read the original abstract

In large language models (LLM)-based recommendation systems (LLM-RSs), accurately predicting user preferences by leveraging the general knowledge of LLMs is possible without requiring extensive training data. By converting recommendation tasks into natural language inputs called prompts, LLM-RSs can efficiently solve issues that have been difficult to address due to data scarcity but are crucial in applications such as cold-start and cross-domain problems. However, when applying this in practice, selecting the prompt that matches tasks and data is essential. Although numerous prompts have been proposed in LLM-RSs and representing the target user in prompts significantly impacts recommendation accuracy, there are still no clear guidelines for selecting specific prompts. In this paper, we categorize and analyze prompts from previous research to establish practical prompt selection guidelines. Through 450 experiments with 90 prompts and five real-world datasets, we examined the relationship between prompts and dataset characteristics in recommendation accuracy. We found that no single prompt consistently outperforms others; thus, selecting prompts on the basis of dataset characteristics is crucial. Here, we propose a prompt selection method that achieves higher accuracy with minimal validation data. Because increasing the number of prompts to explore raises costs, we also introduce a cost-efficient strategy using high-performance and cost-efficient LLMs, significantly reducing exploration costs while maintaining high prediction accuracy. Our work offers valuable insights into the prompt selection, advancing accurate and efficient LLM-RSs.

Figures

Figures reproduced from arXiv: 2412.14454 by the authors.

Figure 1
Figure 1. An example of a prompt for LLM-RSs. We conduct experiments by varying the user’s information part, where the differences from related works are most noticeable. A typical prompt for LLM-RSs is: “The target user has viewed {item1, ..., item10} in the past. Please rank the recommended candidates {item101, ..., item120} in the order of the user’s likely preference” ( [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Summarization prompt, its output text, and its inference prompt. 4 Experiments 4.1 Settings Dataset The data used in the experiments were selected from the Amazon Review Dataset [19], specifically from the categories of Music, Movies, Books, Grocery, and Clothes. For each dataset, 100 users who had rated 30 or more items with a rating of three stars or higher were selected. The auxiliary information for the items in… view at source ↗
Figure 3
Figure 3. Calculation of the relative performance indicator. Using the sampling items approach, we calculated the RPI for 72 prompts ( [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. AutoData: A Multi-Agent System for Open Web Data Collection

    cs.IR 2025-05 conditional novelty 6.0 of 10

    AutoData, a multi-agent system with a hypergraph message cache, automates web dataset collection from a sentence instruction and outperforms general agent baselines on the new Instruct2DS benchmark.

  2. CTG-Insight: A Multi-Agent Interpretable LLM Framework for Cardiotocography Analysis and Classification

    cs.LG 2025-07 conditional novelty 5.0 of 10

    CTG-Insight reports 96.4% accuracy on a 50-sample subset of the NeuroFetalNet test set, but the comparison against deep learning baselines is uneven and the system code and exact prompts are not released.

  3. Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets

    cs.CR 2025-08 unverdicted novelty 4.0 of 10

    Claims two forms of Polymarket arbitrage and $40 million extracted profit, but the body text provided is an unrelated plasma physics paper.

  4. Fine-tuning on simulated data outperforms prompting for agent tone of voice

    cs.LG 2025-07 conditional novelty 4.0 of 10

    Fine-tuning a 1B-parameter LLM on as few as 100 synthetically generated, readability-filtered samples achieved conversational tone more reliably than a verbose system prompt.

Reference graph

Works this paper leans on

38 extracted references · 29 canonical work pages · cited by 4 Pith papers

  1. [1]

    In: RecSys

    Covington, P., Adams, J., Sargin, E.: Deep neural networks for youtube recom- mendations. In: RecSys. pp. 191–198. ACM (2016)

    work page 2016

  2. [2]

    In: RecSys

    Dacrema, M.F., Cremonesi, P., Jannach, D.: Are we really making much progress? A worrying analysis of recent neural recommendation approaches. In: RecSys. pp. 101–109. ACM (2019)

    work page 2019

  3. [3]

    Dai, S., Shao, N., Zhao, H., Yu, W., Si, Z., Xu, C., Sun, Z., Zhang, X., Xu, J.: UncoveringChatGPT’scapabilitiesinrecommendersystems.In:RecSys.pp.1126–

    work page

  4. [4]

    Deldjoo, Y., He, Z., McAuley, J.J., Korikov, A., Sanner, S., Ramisa, A., Vidal, R., Sathiamoorthy, M., Kasirzadeh, A., Milano, S.: A review of modern recom- mender systems using generative models (gen-recsys). In: KDD. pp. 6448–6458. ACM (2024)

    work page 2024

  5. [5]

    In: NAACL-HLT (1)

    Devlin, J., Chang, M., Lee, K., Toutanova, K.: BERT: pre-training of deep bidirec- tional transformers for language understanding. In: NAACL-HLT (1). pp. 4171–

    work page

  6. [6]

    CoRR abs/2309.03613 (2023)

    Di Palma, D., Biancofiore, G.M., Anelli, V.W., Narducci, F., Noia, T.D., Sciascio, E.D.: Evaluating chatgpt as a recommender system: A rigorous approach. CoRR abs/2309.03613 (2023)

    arXiv 2023

  7. [7]

    TKDE01, 1–20 (5555)

    Fan,W.,Zhao,Z.,Li,J.,Liu,Y.,Mei,X.,Wang,Y.,Tang,J.,Li,Q.:Recommender systems in the era of large language models (LLMs). TKDE01, 1–20 (5555)

    work page

  8. [8]

    In: EMNLP (1)

    Gao, T., Yao, X., Chen, D.: Simcse: Simple contrastive learning of sentence embed- dings. In: EMNLP (1). pp. 6894–6910. Association for Computational Linguistics (2021)

    work page 2021

Show all 38 references
  1. [9]

    In: CIKM

    He, Z., Xie, Z., Jha, R., Steck, H., Liang, D., Feng, Y., Majumder, B.P., Kallus, N., McAuley, J.J.: Large language models as zero-shot conversational recommenders. In: CIKM. pp. 720–730. ACM (2023)

    2023

  2. [10]

    In: ECIR (2)

    Hou, Y., Zhang, J., Lin, Z., Lu, H., Xie, R., McAuley, J.J., Zhao, W.X.: Large language models are zero-shot rankers for recommender systems. In: ECIR (2). Lecture Notes in Computer Science, vol. 14609, pp. 364–381. Springer (2024)

    2024

  3. [11]

    In: ECIR (3)

    Ji, J., Li, Z., Xu, S., Hua, W., Ge, Y., Tan, J., Zhang, Y.: Genrec: Large language model for generative recommendation. In: ECIR (3). Lecture Notes in Computer Science, vol. 14610, pp. 494–502. Springer (2024)

    2024

  4. [12]

    ACM Comput

    Khan, M.M., Ibrahim, R., Ghani, I.: Cross domain recommender systems: A sys- tematic literature review. ACM Comput. Surv.50(3), 36:1–36:34 (2017)

    2017

  5. [13]

    Computer42(8), 30–37 (2009)

    Koren, Y., Bell, R.M., Volinsky, C.: Matrix factorization techniques for recom- mender systems. Computer42(8), 30–37 (2009)

    2009

  6. [14]

    In: LREC/COLING

    Li, L., Zhang, Y., Liu, D., Chen, L.: Large language models for generative recom- mendation: A survey and visionary discussions. In: LREC/COLING. pp. 10146– 10159. ELRA and ICCL (2024)

    2024

  7. [15]

    ACM Trans

    Lin, J., Dai, X., Xi, Y., Liu, W., Chen, B., Li, X., Zhu, C., Guo, H., Yu, Y., Tang, R., Zhang, W.: How can recommender systems benefit from large language models: A survey. ACM Trans. Inf. Syst. (2024), just Accepted

    2024

  8. [16]

    Lin, J., Shan, R., Zhu, C., Du, K., Chen, B., Quan, S., Tang, R., Yu, Y., Zhang, W.: Rella: Retrieval-enhanced large language models for lifelong sequential behavior comprehension in recommendation. In: WWW. pp. 3497–3508. ACM (2024)

    2024

  9. [17]

    Liu, J., Shen, D., Zhang, Y., Dolan, B., Carin, L., Chen, W.: What makes good in-context examples for gpt-3? In: DeeLIO@ACL. pp. 100–114. ACL (2022) 14 Kusano at al

    2022

  10. [18]

    CoRRabs/2304.10149 (2023)

    Liu, J., Liu, C., Lv, R., Zhou, K., Zhang, Y.: Is chatgpt a good recommender? A preliminary study. CoRRabs/2304.10149 (2023)

    2023 arXiv

  11. [19]

    In: EMNLP/IJCNLP (1)

    Ni, J., Li, J., McAuley, J.J.: Justifying recommendations using distantly-labeled re- views and fine-grained aspects. In: EMNLP/IJCNLP (1). pp. 188–197. Association for Computational Linguistics (2019)

    2019

  12. [20]

    In: NAACL-HLT

    Pezeshkpour, P., Hruschka, E.: Large language models sensitivity to the order of options in multiple-choice questions. In: NAACL-HLT. pp. 2006–2017. ACL (2024)

    2024

  13. [21]

    org/CorpusID:160025533

    Radford, A., Wu, J., Child, R., Luan, D., Amodei, D., Sutskever, I.: Language mod- els are unsupervised multitask learners (2019), https://api.semanticscholar. org/CorpusID:160025533

    2019

  14. [22]

    CoRR abs/2402.07927 (2024)

    Sahoo, P., Singh, A.K., Saha, S., Jain, V., Mondal, S., Chadha, A.: A systematic survey of prompt engineering in large language models: Techniques and applica- tions. CoRR abs/2402.07927 (2024)

    2024 arXiv

  15. [23]

    In: RecSys

    Sanner, S., Balog, K., Radlinski, F., Wedin, B., Dixon, L.: Large language mod- els are competitive near cold-start recommenders for language- and item-based preferences. In: RecSys. pp. 890–896. ACM (2023)

    2023

  16. [24]

    CoRR abs/2310.14304 (2023)

    Tang, Z., Huan, Z., Li, Z., Zhang, X., Hu, J., Fu, C., Zhou, J., Li, C.: One model for all: Large language models are domain-agnostic recommendation systems. CoRR abs/2310.14304 (2023)

    2023 arXiv

  17. [25]

    CoRR abs/2302.13971 (2023)

    Touvron, H., Lavril, T., Izacard, G., Martinet, X., Lachaux, M., Lacroix, T., Roz- ière, B., Goyal, N., Hambro, E., Azhar, F., Rodriguez, A., Joulin, A., Grave, E., Lample, G.: LLaMA: Open and efficient foundation language models. CoRR abs/2302.13971 (2023)

    2023 arXiv

  18. [26]

    CoRRabs/2407.12994 (2024)

    Vatsal, S., Dubey, H.: A survey of prompt engineering methods in large language models for different nlp tasks. CoRRabs/2407.12994 (2024)

    2024 arXiv

  19. [27]

    In: NeurIPS

    Volkovs, M., Yu, G.W., Poutanen, T.: Dropoutnet: Addressing cold start in rec- ommender systems. In: NeurIPS. pp. 4957–4966 (2017)

    2017

  20. [28]

    In: WWW (Companion Volume)

    Wang, J., Lu, H., Caverlee, J., Chi, E.H., Chen, M.: Large language models as data augmenters for cold-start item recommendation. In: WWW (Companion Volume). pp. 726–729. ACM (2024)

    2024

  21. [29]

    CoRRabs/2304.03153 (2023)

    Wang, L., Lim, E.: Zero-shot next-item recommendation using large pretrained language models. CoRRabs/2304.03153 (2023)

    2023 arXiv

  22. [30]

    In: NAACL-HLT

    Wang, L., Lim, E.: The whole is better than the sum: Using aggregated demon- strations in in-context learning for sequential recommendation. In: NAACL-HLT. pp. 876–895. ACL (2024)

    2024

  23. [31]

    In: NAACL-HLT

    Wang, Y., Jiang, Z., Chen, Z., Yang, F., Zhou, Y., Cho, E., Fan, X., Lu, Y., Huang, X., Yang, Y.: Recmind: Large language model powered agent for recommendation. In: NAACL-HLT. pp. 4351–4364. ACL (2024)

    2024

  24. [32]

    In: ACM Multimedia

    Wei, Y., Wang, X., Li, Q., Nie, L., Li, Y., Li, X., Chua, T.: Contrastive learning for cold-start recommendation. In: ACM Multimedia. pp. 5382–5390. ACM (2021)

    2021

  25. [33]

    World Wide Web (WWW)27(5), 60 (2024)

    Wu, L., Zheng, Z., Qiu, Z., Wang, H., Gu, H., Shen, T., Qin, C., Zhu, C., Zhu, H., Liu, Q., Xiong, H., Chen, E.: A survey on large language models for recommenda- tion. World Wide Web (WWW)27(5), 60 (2024)

    2024

  26. [34]

    Wu, X., Zhou, H., Shi, Y., Yao, W., Huang, X., Liu, N.: Could small language models serve as recommenders? towards data-centric cold-start recommendation. In: WWW. pp. 3566–3575. ACM (2024)

    2024

  27. [35]

    CoRRabs/2401.04997 (2024) On Prompt Selection in LLM-RSs 15

    Xu, L., Zhang, J., Li, B., Wang, J., Cai, M., Zhao, W.X., Wen, J.: Prompting large language models for recommender systems: A comprehensive framework and empirical analysis. CoRRabs/2401.04997 (2024) On Prompt Selection in LLM-RSs 15

    2024 arXiv

  28. [36]

    Yue, Z., Rabhi, S., de Souza Pereira Moreira, G., Wang, D., Oldridge, E.: Lla- maRec:Two-stagerecommendationusinglargelanguagemodelsforranking.CoRR abs/2311.02089 (2023)

    2023 arXiv

  29. [37]

    Zhu, Y., Wu, L., Guo, Q., Hong, L., Li, J.: Collaborative large language model for recommender systems. In: WWW. pp. 3162–3172. ACM (2024)

    2024

  30. [38]

    In: WSDM

    Zhu, Y., Tang, Z., Liu, Y., Zhuang, F., Xie, R., Zhang, X., Lin, L., He, Q.: Person- alized transfer of user preferences for cross-domain recommendation. In: WSDM. pp. 1507–1515. ACM (2022)

    2022

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.