Pith. sign in

REVIEW 5 major objections 6 minor 91 references

A two-stage language-model recommender lifts grocery cart-adds per impression by 7.5% over the production system.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-02 12:10 UTC pith:4FV4G5IN

load-bearing objection A practical two-stage LoRA plus trie-decode recipe for grocery category recommendation with a plausible but under-specified production lift; the main fixable flaw is the undisclosed split for the rebuying statistics in Eq. (2). the 5 major comments →

arxiv 2607.24764 v1 pith:4FV4G5IN submitted 2026-06-04 cs.AI

GrocLM: Grocery Category Recommendation in E-Commerce with Large Language Models

Yuan Zhong , Chuanwei Ruan , Moein Hasani , Tejaswi Tenneti , Haixun Wang , Fenglong Ma This is my paper
classification cs.AI
keywords grocery recommendationcategory-level recommendationLLM fine-tuningLoRArebuying cycletrie-constrained decodinggenerative retrievale-commerce
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Grocery purchases are cyclical: shoppers rebuy staples on predictable cycles and build baskets across complementary categories. GrocLM is a fine-tuned language model that treats grocery category recommendation as constrained text generation, encoding each category's typical repurchase gap directly into the model's parameters rather than into the prompt. A second tuning stage teaches it which categories co-occur given current search queries and past conversions, and a trie-shaped output constraint guarantees the model only names valid categories. On proprietary production data the model beats both conventional retrieval baselines and LLM-based recommenders across Precision@K, Recall@K, and F1@K, and in a live restocking task it improves cart-adds per impression by 7.5% relative to the deployed model. If correct, this is evidence that a single autoregressive pass over a category vocabulary can replace per-category scoring pipelines.

Core claim

The central claim is that grocery category recommendation can be recast as generative language modeling and outperform existing pipelines at production scale. The model first learns unconditional category-level repurchase cycles—the mean and standard deviation of days between buys—through a LoRA fine-tuning stage, then learns conditional category relationships from sequences of past conversions and current queries. Output validity is guaranteed by a trie-based masking scheme that only permits tokens that continue a legitimate category name. The paper reports that this system beats all tested baselines on the proprietary Conversion dataset and on the public Aisle benchmark, and that in a live

What carries the argument

The argument rides on three linked mechanisms. First, category-level rebuying statistics—for each category, the mean and standard deviation of the time between a user's repeat purchases, averaged over users—are injected via LoRA fine-tuning, so the cyclical prior lives in the model weights rather than in the prompt. Second, relational fine-tuning conditions the model on the user's current queries and past conversion categories to capture in-session intent and category co-occurrence. Third, category names are organized into a trie, and beam-search decoding masks any token that would leave the trie, so the model can only emit valid, mutually exclusive category labels. The trie is dynamic: in p

Load-bearing premise

The load-bearing premise is that the Stage-1 rebuying statistics are computed only from the training portion of the conversion data, so no signal from the evaluation period leaks into the model; the production 7.5% also assumes the A/B comparison is a clean, like-for-like test against the incumbent system.

What would settle it

Recompute the Stage-1 category rebuying statistics (mean and std of purchase gaps) using only timestamps that strictly precede the evaluation period and rerun the offline tables; if the margin over baselines collapses, the leakage concern is real. For the production claim, request the A/B setup with the incumbent model and a confidence interval on cart-adds per impression; if the interval brackets zero, the claimed lift is not established.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • A single fine-tuned language model can handle category-level recommendation jointly, removing the need for separate per-category scoring models.
  • Parameter-encoded rebuying cycles (LoRA) beat the same statistics supplied as in-context prompt text by a wide margin, suggesting this is a more effective injection path for cyclical priors.
  • Removing trie-constrained decoding costs 5.1% in production cart-adds, so output validity is not a minor detail—it is a principal source of the gain.
  • Joint generation in about 0.3 seconds per session is fast enough for large-scale commercial deployment, and the event-driven streaming design points to sub-second updates.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • My inference: the dynamic per-user trie used in production is a strong prior by itself—restricting output to previously purchased categories—so part of the 7.5% lift may be attributable to that restriction rather than to the language model's learned rankings; a controlled experiment with a fixed vocabulary would separate the two.
  • My inference: the two-stage recipe is a template for any domain with periodic demand and a fixed label set, such as fashion (seasonal cycles) or pet supplies (replenishment cycles); the paper lists this extension but does not test it.
  • My inference: the cold-start result that an empty input produces a static 'Baking Supplies' output shows the model has learned a global popularity prior; adding demographic or session-metadata signals, which the paper flags as future work, is the natural next test for sparse-user robustness.
  • My inference: because the paper does not report a confidence interval for the 7.5% production figure, a fair reader should treat it as a point estimate from one deployment window rather than as a guaranteed effect size.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. GrocLM proposes a two-stage LoRA fine-tuning strategy for LLM-based grocery category recommendation. In Stage 1, the model is fine-tuned to output category-level rebuying statistics (mean and standard deviation of inter-purchase time gaps) computed from conversion histories; in Stage 2, it is fine-tuned to predict the set of categories a user will convert to given past conversions and current queries. A trie-based constrained decoding mechanism with beam search guarantees valid category labels. Experiments on a proprietary Conversion dataset and the public Instacart Aisle dataset report consistent improvements over a range of traditional and LLM-based baselines on Precision@K, Recall@K, and F1@K; a production restocking task reports a 7.5% relative improvement in cart-adds per impression over the deployed model.

Significance. If fully supported, the paper provides a concrete industrial validation that a two-stage LoRA-tuned LLM with constrained decoding can outperform an incumbent production system for category-level grocery recommendation. The work is potentially valuable for practitioners because it addresses an underexplored task (category-level rather than item-level recommendation), explicitly models cyclical repurchasing, and shows a principled way to constrain LLM outputs to a fixed label space. Strengths include the inclusion of a public benchmark, ablations, case studies, data-size sensitivity analysis, and honest discussion of limitations in Appendices E and G. However, the empirical support currently has gaps: missing standard deviations, an unspecified split for the rebuying statistics, and an under-specified production experiment. These gaps are load-bearing for the paper's central claims of consistent offline superiority and of a measurable production lift.

major comments (5)
  1. [§4.1, Eq. (2)] The Rebuying Statistics dataset is constructed from the Conversion dataset by computing the mean and standard deviation of purchase-time gaps per category. The paper never states whether this computation is restricted to the training portion of the data. Since Stage 1 fine-tuning trains the model to output these (µ_c, σ_c), computing them over the full six-month window would inject aggregate future demand information into the model, biasing all offline results (Tables 2, 4, 5, and the ablation in Table 3). Please clarify the split used for Eq. (2); if the statistics are computed on the full window, recompute with training-only data and reassess the offline comparisons.
  2. [§4.5, Figure 5] The 7.5% production lift is the headline practical result, but the experiment is described in only a few sentences. No information is given about the deployed production model's architecture, training data, or update frequency; the experiment's duration and traffic allocation; or whether the difference is statistically significant. The comparison is between GrocLM with a user-specific trie and an unspecified production model, making the result impossible to interpret. Please provide a complete A/B experiment design, including confidence intervals and a description of the production baseline, or temper the production claim accordingly.
  3. [§4.3, Table 2; Appendix A, Table 5] The text states that 'we report the 5-run mean and standard deviation,' but Tables 2 and 5 contain only single values per cell. Without standard deviations, the claim of consistent outperformance cannot be evaluated statistically. Please add the standard deviations (or a supplementary table) and clarify what varies across the five runs (random seeds, user subsamples, etc.).
  4. [§4.4, Table 3 vs §4.5] Ablation AS4 (without constrained decoding) yields exactly 0.000 on every offline metric, while removing constrained decoding in production results in only a 5.1% relative regression. This large discrepancy suggests that the offline metric is dominated by exact token/string matching: any surface-form deviation is counted as a complete miss. Reporting only exact-match Precision/Recall therefore conflates output validity with recommendation quality and may exaggerate the benefit of the trie. Please report a variant where unconstrained outputs are aligned to the category vocabulary (e.g., via fuzzy matching or by post-processing through the trie) and discuss the difference.
  5. [§4.1] The train/test split for the Conversion dataset is not described. It is unclear whether the split is temporal (e.g., last week held out) or random, and whether the same users appear in both training and evaluation. In a setting with cyclical repurchasing, this distinction is essential for assessing temporal leakage. Please state the split explicitly and confirm that the Rebuying Statistics dataset (Eq. 2) is computed using only the training portion.
minor comments (6)
  1. [§4.3] The evaluation metrics are described as Precision@K/Recall@K/F1@K over category sets, yet the text also says 'ranking is not considered.' Please define precisely how the top-K set is selected from the generated output (e.g., generation order, beam order, or sampling). Without this, the metrics are ambiguous.
  2. [§4.1, Table 1] Table 1 reports 'Distinct Orders 32,14,874' which appears to be a typo; please correct the number formatting.
  3. [§4.6] The ICL-only variant is described only briefly. Please provide the exact prompt template used for the in-context learning comparison so that the result in Table 4 is reproducible.
  4. [§3.3] The statement that 'new categories can be incorporated by updating the trie without retraining the model' overstates the case: the model's output distribution was trained on the existing category vocabulary, so adding a new category to the trie does not mean the model has learned to generate it. Please revise to avoid implying zero-cost generalization to unseen categories.
  5. [Figure 5] The y-axis label for Figure 5 is missing. Please specify whether the plotted values are relative improvements or absolute metric values, and state the units clearly.
  6. [Appendix E] The cold-start limitation—the model outputs a static 'Baking Supplies' when neither history nor query is provided—is openly acknowledged. Consider moving this discussion into the main text or at least cross-referencing it in the conclusion, since it is directly relevant to the claimed generality.

Circularity Check

0 steps flagged

No significant circularity; core recommendation claims are evaluated on held-out behavior independent of Stage-1 cycle labels.

full rationale

The central claim—offline tables 2 and 5 plus the production 7.5% lift—rests on predicting held-out category sets I_M from history I and queries Q. Stage 1's Rebuying Statistics task fine-tunes the LLM to reproduce precomputed (μ_c, σ_c); this is an auxiliary pretraining objective, not the evaluated output, and the final predictor never computes I_M from μ_c, σ_c. Stage 2 fine-tunes on sequence data, and evaluation uses ground-truth conversion categories. Eq. (1) and Eq. (4) are standard autoregressive likelihood; Eq. (3) restricts decoding to the vocabulary trie but does not define the target. No load-bearing step is justified solely by a self-citation, and the public Instacart Aisle benchmark in Appendix A provides an external check. The undisclosed train/eval split for Eq. (2) is a data-leakage/correctness risk, not circularity, because it does not make the evaluation quantity equal to the training objective by construction.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 0 invented entities

The central claim depends on the standard empirical scaffolding of the paper: data filters, LoRA hyperparameters, and especially the data-derived rebuying statistics injected in Stage 1. All of these are fitted or chosen from the same proprietary domain; none is independently verified outside the paper. The domain assumptions are that purchase intervals are summarized by mean/std, that aggregate statistics are stable over time, and that the fixed category vocabulary is complete.

free parameters (4)
  • per-category rebuying statistics (µ_c, σ_c) = computed from timestamps; e.g., banana (7.3, 1.0) in Fig. 2
    Used as Stage-1 supervision; these aggregate purchase-interval statistics are fitted from the proprietary conversion data.
  • LoRA rank and alpha = r=64, α=16, dropout 0.1
    Hand-chosen hyperparameters for both fine-tuning stages.
  • user activity filters = ≥5 conversions in 6 months; ≥10 total purchases
    Defines the Conversion dataset and excludes low-activity users from the main evaluation.
  • category support threshold = ≥5 users with repeat purchases
    Hand-chosen cutoff for including a category's rebuying statistics.
axioms (5)
  • domain assumption Autoregressive factorization of the category-set probability (Eq. 1) treats the set I_M as a token sequence despite the categories being independent and mutually exclusive.
    The model is trained with a language-model objective over an ordered list of category names; the ordering is an arbitrary serialization of a set.
  • domain assumption Rebuying intervals are adequately summarized by mean and standard deviation, and Eq. (2) aggregates per-user means without weights.
    Stage 1 uses only (µ_c, σ_c) as the cyclical signal; if the true gap distribution is multimodal or heavy-tailed, this summary loses information.
  • domain assumption Aggregate rebuying statistics computed from historical conversions remain valid for future prediction windows (stationarity).
    The model is trained and evaluated on data from overlapping or successive time windows; the paper does not analyze drift.
  • domain assumption The trie over the predefined category space contains every valid output, and new categories can be added by updating the trie without retraining.
    Section 3.3 states this; if the category space is incomplete, constrained decoding will force false negatives.
  • domain assumption LLAMA-3-8B-Instruct is an appropriate base model and LoRA fine-tuning is sufficient to adapt it to grocery category prediction.
    The choice of base model is a design decision, not derived from first principles.

pith-pipeline@v1.3.0-alltime-deepseek · 16839 in / 16542 out tokens · 164486 ms · 2026-08-02T12:10:44.077781+00:00 · methodology

0 comments
read the original abstract

The rapid growth of online grocery shopping requires recommendation systems that capture cyclical purchasing behavior and diverse user intents. Traditional item-level methods face scalability and accuracy challenges, motivating category-level recommendation as a more structured and practical alternative. We present GROCLM, a fine-tuned language model for grocery category recommendation in a real-world production environment. GROCLM employs a two-stage LoRA-based training strategy to encode cyclical purchasing patterns directly into model parameters, enabling more effective utilization of rebuying signals compared to prompt-based conditioning. To ensure valid and controllable outputs, we further introduce a trie-based constrained decoding mechanism over a predefined category space. Experiments on both proprietary production data and a public benchmark demonstrate that GROCLM consistently outperforms strong baselines. In a live production restocking task, GROCLM achieves a 7.5% relative improvement in cart-adds per impression, while maintaining efficient inference by generating all categories jointly. These results highlight the effectiveness and practicality of integrating large language models into structured recommendation systems.

Figures

Figures reproduced from arXiv: 2607.24764 by Chuanwei Ruan, Fenglong Ma, Haixun Wang, Moein Hasani, Tejaswi Tenneti, Yuan Zhong.

Figure 1
Figure 1. Figure 1: Illustration of grocery category recommen [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed GROCLM item relationships. The emergence of transformer￾based models (Vaswani, 2017) and BERT (Devlin, 2018) further enhanced contextualized represen￾tation learning. Bi-encoder methods (Karpukhin et al., 2020; Qu et al., 2020; Xiong et al., 2020; Ni et al., 2021a,b; Reimers and Gurevych, 2019) and multi-representation models (Humeau et al., 2019; Luo et al., 2022; Khattab and Zaha… view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of trie tree. categories can be incorporated by updating the trie without retraining the model. 4 Experiments In this section, we compare our model’s perfor￾mance on various metrics, datasets, and prediction tasks. 4.1 Datasets We evaluate GROCLM on a proprietary large-scale grocery dataset, referred to as the Conversion dataset. Statistics of the Conversion dataset are shown in [PITH_FULL_IM… view at source ↗
Figure 4
Figure 4. Figure 4: Data size sensitivity analysis. From Left to right: Recall@10, Precision@10, and F1@10. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Relative improvement of GROCLM over the production model. (w/o Cons) refers to the variant with￾out constrained generation. Model P@5 R@5 F1@5 P@10 R@10 F1@10 P@20 R@20 F1@20 ICL Only 0.010 0.008 0.008 0.006 0.012 0.007 0.004 0.016 0.006 GROCLM 0.142 0.186 0.126 0.094 0.231 0.111 0.069 0.284 0.097 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Illustration of the case study with a user’s buying history. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of the case study of the rebuying cycle. [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

91 extracted references · 30 linked inside Pith

  1. [1]

    Advances in Neural Information Processing Systems , volume=

    Transformer memory as a differentiable search index , author=. Advances in Neural Information Processing Systems , volume=

  2. [2]

    arXiv preprint arXiv:2010.00904 , year=

    Autoregressive entity retrieval , author=. arXiv preprint arXiv:2010.00904 , year=

    Pith/arXiv arXiv 2010

  3. [3]

    arXiv preprint arXiv:2110.07602 , year=

    P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks , author=. arXiv preprint arXiv:2110.07602 , year=

    Pith/arXiv arXiv

  4. [4]

    Advances in neural information processing systems , volume=

    Large language models are zero-shot reasoners , author=. Advances in neural information processing systems , volume=

  5. [5]

    Advances in Neural Information Processing Systems , volume=

    Generating training data with language models: Towards zero-shot language understanding , author=. Advances in Neural Information Processing Systems , volume=

  6. [6]

    ACM Computing Surveys , volume=

    Recent advances in natural language processing via large pre-trained language models: A survey , author=. ACM Computing Surveys , volume=. 2023 , publisher=

    2023

  7. [7]

    arXiv preprint arXiv:2302.13971 , year=

    Llama: Open and efficient foundation language models , author=. arXiv preprint arXiv:2302.13971 , year=

    Pith/arXiv arXiv

  8. [8]

    arXiv preprint arXiv:2303.08774 , year=

    Gpt-4 technical report , author=. arXiv preprint arXiv:2303.08774 , year=

    Pith/arXiv arXiv

  9. [9]

    arXiv preprint arXiv:1908.10084 , year=

    Sentence-bert: Sentence embeddings using siamese bert-networks , author=. arXiv preprint arXiv:1908.10084 , year=

    Pith/arXiv arXiv 1908

  10. [10]

    Proceedings of the 22nd ACM international conference on Information & Knowledge Management , pages=

    Learning deep structured semantic models for web search using clickthrough data , author=. Proceedings of the 22nd ACM international conference on Information & Knowledge Management , pages=

  11. [11]

    2018 IEEE international conference on data mining (ICDM) , pages=

    Self-attentive sequential recommendation , author=. 2018 IEEE international conference on data mining (ICDM) , pages=. 2018 , organization=

    2018

  12. [12]

    Advances in Neural Information Processing Systems , volume=

    Recommender systems with generative retrieval , author=. Advances in Neural Information Processing Systems , volume=

  13. [13]

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    On generative agents in recommendation , author=. Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  14. [14]

    arXiv preprint arXiv:2404.16924 , year=

    A survey of generative search and recommendation in the era of large language models , author=. arXiv preprint arXiv:2404.16924 , year=

    Pith/arXiv arXiv

  15. [15]

    Proceedings of the first instructional conference on machine learning , volume=

    Using tf-idf to determine word relevance in document queries , author=. Proceedings of the first instructional conference on machine learning , volume=. 2003 , organization=

    2003

  16. [16]

    Proceedings of the 16th ACM Conference on Recommender Systems , pages=

    Recommendation as language processing (rlp): A unified pretrain, personalized prompt & predict paradigm (p5) , author=. Proceedings of the 16th ACM Conference on Recommender Systems , pages=

  17. [17]

    arXiv preprint arXiv:2304.03516 , year=

    Generative recommendation: Towards next-generation recommender paradigm , author=. arXiv preprint arXiv:2304.03516 , year=

    Pith/arXiv arXiv

  18. [18]

    2024 IEEE 40th International Conference on Data Engineering (ICDE) , pages=

    Adapting large language models by integrating collaborative semantics for recommendation , author=. 2024 IEEE 40th International Conference on Data Engineering (ICDE) , pages=. 2024 , organization=

    2024

  19. [19]

    arXiv preprint arXiv:2403.19021 , year=

    Towards llm-recsys alignment with textual id learning , author=. arXiv preprint arXiv:2403.19021 , year=

    Pith/arXiv arXiv

  20. [20]

    arXiv preprint arXiv:2309.13375 , year=

    Generative Retrieval with Semantic Tree-Structured Item Identifiers via Contrastive Learning , author=. arXiv preprint arXiv:2309.13375 , year=

    Pith/arXiv arXiv

  21. [21]

    arXiv preprint arXiv:2403.18480 , year=

    Enhanced generative recommendation via content and collaboration integration , author=. arXiv preprint arXiv:2403.18480 , year=

    Pith/arXiv arXiv

  22. [22]

    Information Processing & Management , volume=

    An information-theoretic perspective of tf--idf measures , author=. Information Processing & Management , volume=. 2003 , publisher=

    2003

  23. [23]

    Foundations and Trends

    The probabilistic relevance framework: BM25 and beyond , author=. Foundations and Trends. 2009 , publisher=

    2009

  24. [24]

    arXiv preprint arXiv:2308.08434 , year=

    A bi-step grounding paradigm for large language models in recommendation systems , author=. arXiv preprint arXiv:2308.08434 , year=

    Pith/arXiv arXiv

  25. [25]

    Proceedings of the ACM on Web Conference 2024 , pages=

    Collaborative large language model for recommender systems , author=. Proceedings of the ACM on Web Conference 2024 , pages=

    2024

  26. [26]

    IEEE Transactions on Knowledge and Data Engineering , year=

    Query-oriented Data Augmentation for Session Search , author=. IEEE Transactions on Knowledge and Data Engineering , year=

  27. [27]

    Proceedings of the 31st ACM International Conference on Information & Knowledge Management , pages=

    Query-aware sequential recommendation , author=. Proceedings of the 31st ACM International Conference on Information & Knowledge Management , pages=

  28. [28]

    arXiv preprint arXiv:2406.10450 , year=

    TokenRec: Learning to Tokenize ID for LLM-based Generative Recommendation , author=. arXiv preprint arXiv:2406.10450 , year=

    Pith/arXiv arXiv

  29. [29]

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    IDGenRec: LLM-RecSys Alignment with Textual ID Learning , author=. Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  30. [30]

    arXiv preprint arXiv:2406.01197 , year=

    A Survey of Generative Information Retrieval , author=. arXiv preprint arXiv:2406.01197 , year=

    Pith/arXiv arXiv

  31. [31]

    Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP) , pages=

    Glove: Global vectors for word representation , author=. Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP) , pages=

    2014

  32. [32]

    arXiv preprint arXiv:2407.21783 , year=

    The llama 3 herd of models , author=. arXiv preprint arXiv:2407.21783 , year=

    Pith/arXiv arXiv

  33. [33]

    Telematics and Informatics , volume=

    Online grocery shopping before and during the COVID-19 pandemic: A meta-analytical review , author=. Telematics and Informatics , volume=. 2022 , publisher=

    2022

  34. [34]

    Plos one , volume=

    The demand for online grocery shopping: COVID-induced changes in grocery shopping behavior of Canadian consumers , author=. Plos one , volume=. 2024 , publisher=

    2024

  35. [35]

    World Wide Web , volume=

    A survey on large language models for recommendation , author=. World Wide Web , volume=. 2024 , publisher=

    2024

  36. [36]

    IEEE Transactions on Knowledge and Data Engineering , year=

    Recommender systems in the era of large language models (llms) , author=. IEEE Transactions on Knowledge and Data Engineering , year=

  37. [37]

    World Wide Web , volume=

    When large language models meet personalization: Perspectives of challenges and opportunities , author=. World Wide Web , volume=. 2024 , publisher=

    2024

  38. [38]

    Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Training large-scale news recommenders with pretrained language models in the loop , author=. Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

  39. [39]

    Proceedings of the 17th ACM Conference on Recommender Systems , pages=

    Tallrec: An effective and efficient tuning framework to align large language model with recommendation , author=. Proceedings of the 17th ACM Conference on Recommender Systems , pages=

  40. [40]

    arXiv preprint arXiv:2205.08084 , year=

    M6-rec: Generative pretrained language models are open-ended recommender systems , author=. arXiv preprint arXiv:2205.08084 , year=

    Pith/arXiv arXiv

  41. [41]

    Proceedings of the Annual International ACM SIGIR Conference on Research and Development in Information Retrieval in the Asia Pacific Region , pages=

    How to index item ids for recommendation foundation models , author=. Proceedings of the Annual International ACM SIGIR Conference on Research and Development in Information Retrieval in the Asia Pacific Region , pages=

  42. [42]

    Proceedings of the 17th ACM Conference on Recommender Systems , pages=

    Uncovering chatgpt’s capabilities in recommender systems , author=. Proceedings of the 17th ACM Conference on Recommender Systems , pages=

  43. [43]

    arXiv preprint arXiv:2305.06474 , year=

    Do llms understand user preferences? evaluating llms on user rating prediction , author=. arXiv preprint arXiv:2305.06474 , year=

    Pith/arXiv arXiv

  44. [44]

    Journal of Theoretical and Applied Electronic Commerce Research , volume=

    Online grocery shopping in Germany: Assessing the impact of COVID-19 , author=. Journal of Theoretical and Applied Electronic Commerce Research , volume=. 2022 , publisher=

    2022

  45. [45]

    Proceedings of the 17th ACM Conference on Recommender Systems , pages=

    Is chatgpt fair for recommendation? evaluating fairness in large language model recommendation , author=. Proceedings of the 17th ACM Conference on Recommender Systems , pages=

  46. [46]

    Proceedings of the 44th international ACM SIGIR conference on research and development in information retrieval , pages=

    Empowering news recommendation with pre-trained language models , author=. Proceedings of the 44th international ACM SIGIR conference on research and development in information retrieval , pages=

  47. [47]

    arXiv preprint arXiv:2106.09685 , year=

    Lora: Low-rank adaptation of large language models , author=. arXiv preprint arXiv:2106.09685 , year=

    Pith/arXiv arXiv

  48. [48]

    InstaCart Online Grocery Basket Analysis Dataset , year =

  49. [49]

    Proceedings of the 17th ACM Conference on Recommender Systems , pages=

    Leveraging large language models for sequential recommendation , author=. Proceedings of the 17th ACM Conference on Recommender Systems , pages=

  50. [50]

    arXiv preprint arXiv:2305.15756 , year=

    Unitrec: A unified text-to-text transformer and joint contrastive learning framework for text-based recommendation , author=. arXiv preprint arXiv:2305.15756 , year=

    Pith/arXiv arXiv

  51. [51]

    Proceedings of the 17th ACM conference on recommender systems , pages=

    Large language models are competitive near cold-start recommenders for language-and item-based preferences , author=. Proceedings of the 17th ACM conference on recommender systems , pages=

  52. [52]

    European Conference on Information Retrieval , pages=

    Zero-shot recommendation as language modeling , author=. European Conference on Information Retrieval , pages=. 2022 , organization=

    2022

  53. [53]

    Proceedings of the 29th ACM International Conference on Information & Knowledge Management , pages=

    Generate neural template explanations for recommendation , author=. Proceedings of the 29th ACM International Conference on Information & Knowledge Management , pages=

  54. [54]

    ACM Transactions on Information Systems , volume=

    Personalized prompt learning for explainable recommendation , author=. ACM Transactions on Information Systems , volume=. 2023 , publisher=

    2023

  55. [55]

    Justifying recommendations using distantly-labeled reviews and fine-grained aspects , author=. Proceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP) , pages=

    2019

  56. [56]

    Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    Rexplug: Explainable recommendation using plug-and-play language model , author=. Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  57. [57]

    arXiv preprint arXiv:2304.10149 , year=

    Is chatgpt a good recommender? a preliminary study , author=. arXiv preprint arXiv:2304.10149 , year=

    Pith/arXiv arXiv

  58. [58]

    arXiv preprint arXiv:2310.19488 , year=

    Collm: Integrating collaborative embeddings into large language models for recommendation , author=. arXiv preprint arXiv:2310.19488 , year=

    Pith/arXiv arXiv

  59. [59]

    Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    A review of modern recommender systems using generative models (gen-recsys) , author=. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

  60. [60]

    Proceedings of the December 4-6, 1962, fall joint computer conference , pages=

    Some experiments in the generation of word and document associations , author=. Proceedings of the December 4-6, 1962, fall joint computer conference , pages=

    1962

  61. [61]

    Communications of the ACM , volume=

    A vector space model for automatic indexing , author=. Communications of the ACM , volume=. 1975 , publisher=

    1975

  62. [62]

    Journal of documentation , volume=

    Understanding inverse document frequency: on theoretical arguments for IDF , author=. Journal of documentation , volume=. 2004 , publisher=

    2004

  63. [63]

    ACM computing surveys (CSUR) , volume=

    Inverted files for text search engines , author=. ACM computing surveys (CSUR) , volume=. 2006 , publisher=

    2006

  64. [64]

    ACM Transactions on Database Systems (TODS) , volume=

    Inverted files versus signature files for text indexing , author=. ACM Transactions on Database Systems (TODS) , volume=. 1998 , publisher=

    1998

  65. [65]

    Foundations and Trends

    Learning to rank for information retrieval , author=. Foundations and Trends. 2009 , publisher=

    2009

  66. [66]

    Proceedings of the 25th ACM international on conference on information and knowledge management , pages=

    A deep relevance matching model for ad-hoc retrieval , author=. Proceedings of the 25th ACM international on conference on information and knowledge management , pages=

  67. [67]

    Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    Matchzoo: A learning, practicing, and developing system for neural text matching , author=. Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  68. [68]

    arXiv preprint arXiv:1705.01509 , year=

    Neural models for information retrieval , author=. arXiv preprint arXiv:1705.01509 , year=

    Pith/arXiv arXiv

  69. [69]

    Advances in Neural Information Processing Systems , year=

    Attention is all you need , author=. Advances in Neural Information Processing Systems , year=

  70. [70]

    arXiv preprint arXiv:1810.04805 , year=

    Bert: Pre-training of deep bidirectional transformers for language understanding , author=. arXiv preprint arXiv:1810.04805 , year=

    Pith/arXiv arXiv

  71. [71]

    ACM Transactions on Information Systems (TOIS) , volume=

    Semantic models for the first-stage retrieval: A comprehensive review , author=. ACM Transactions on Information Systems (TOIS) , volume=. 2022 , publisher=

    2022

  72. [72]

    Foundations and Trends

    Pre-training methods in information retrieval , author=. Foundations and Trends. 2022 , publisher=

    2022

  73. [73]

    Proceedings of the 14th ACM International Conference on web search and data mining , pages=

    Pretrained transformers for text ranking: BERT and beyond , author=. Proceedings of the 14th ACM International Conference on web search and data mining , pages=

  74. [74]

    arXiv preprint arXiv:2004.04906 , year=

    Dense passage retrieval for open-domain question answering , author=. arXiv preprint arXiv:2004.04906 , year=

    Pith/arXiv arXiv 2004

  75. [75]

    arXiv preprint arXiv:2010.08191 , year=

    RocketQA: An optimized training approach to dense passage retrieval for open-domain question answering , author=. arXiv preprint arXiv:2010.08191 , year=

    Pith/arXiv arXiv 2010

  76. [76]

    arXiv preprint arXiv:2007.00808 , year=

    Approximate nearest neighbor negative contrastive learning for dense text retrieval , author=. arXiv preprint arXiv:2007.00808 , year=

    Pith/arXiv arXiv 2007

  77. [77]

    arXiv preprint arXiv:2108.08877 , year=

    Sentence-t5: Scalable sentence encoders from pre-trained text-to-text models , author=. arXiv preprint arXiv:2108.08877 , year=

    Pith/arXiv arXiv

  78. [78]

    arXiv preprint arXiv:2112.07899 , year=

    Large dual encoders are generalizable retrievers , author=. arXiv preprint arXiv:2112.07899 , year=

    Pith/arXiv arXiv

  79. [79]

    arXiv preprint arXiv:1905.01969 , year=

    Poly-encoders: Transformer architectures and pre-training strategies for fast and accurate multi-sentence scoring , author=. arXiv preprint arXiv:1905.01969 , year=

    Pith/arXiv arXiv 1905

  80. [80]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Improving biomedical information retrieval with neural retrievers , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

Showing first 80 references.