Pith. sign in

REVIEW 4 major objections 6 minor 36 references

Rethinking negative sampling in content-based news recommendation

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper claims that sampling the news articles farthest from a user's reading history as negative examples lets a simple per-user network match state-of-the-art news recommenders.

desk verdict Useful negative-sampling idea with a clean per-user framing, but the offline evaluation is currently undermined by a temporal-leakage risk in the sampler and a weak SOTA comparison. read the letter →

arxiv 2411.08700 v1 pith:CHKO7KZC submitted 2024-11-13 cs.IR cs.AIcs.LG

Miguel Ângelo Rebelo , João Vinagre , Ivo Pereira , Álvaro Figueira This is my paper · ORCID
classification cs.IRcs.AIcs.LG
keywords negativesamplingnewsrecommendationcontent-baseddecentralizedsentenceembeddingsper-usermodelsMINDdataset
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 argues that in content-based news recommendation, how negative examples are chosen matters more than model size. It proposes a per-user recommender whose training set pairs the user's clicked articles with an equal number of articles deliberately chosen as the farthest from the user's average reading history in a sentence-embedding space. A small feed-forward network trained on this balanced 'synthetic pool' reaches AUC values comparable to much larger state-of-the-art models on the MIND dataset, while training much faster and with far fewer parameters. Because each user has their own small network, the approach is naturally suited to on-device or edge training, which the paper connects to privacy, latency, and scalability benefits.

What carries the argument

The load-bearing mechanism is the synthetic negative sampler (Algorithm 1). Title embeddings are L2-normalized, a per-user centroid is computed as the mean of the user's clicked-article embeddings, and all news embeddings are ranked by inner product with that centroid; because the vectors are normalized, squared Euclidean distance equals $2 - 2\cos(x,y)$, so the largest distances correspond to the smallest cosine similarities. The sampler takes the $n$ farthest articles as the user's synthetic negatives, with $n$ equal to the number of clicked articles and capped at the 60 most recent ones, producing a perfectly balanced training set. This mechanism is what lets a simple feed-forward net separate read from uninteresting news cleanly, as the paper's PCA comparison shows.

What would settle it

A direct check would be to take a user's held-out later clicks and ask whether those clicked articles rank among the items the sampler labels negative because they are farthest from the user's earlier reading centroid; if a meaningful share of later clicks are condemned as negatives, the labeling assumption is wrong. A second concrete test is to retrain with a strict time-based split, drawing negatives only from articles published before the test impressions, and see whether the AUC advantage over random sampling survives.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central result is that a lightweight content-based news recommender can be competitive with the state of the art without a complex architecture, provided the negative labels are informative. The DNNR method embeds news titles with a multilingual MiniLM model, concatenates category and type one-hot vectors, and for each user trains a tiny feed-forward network on the user's recent clicked articles plus an equal number of synthetic negatives: the news articles whose embeddings are farthest from the user's embedding centroid. On the MIND-small benchmark, this approach produces mean AUC close to NRMS, DKN, and NPA, and it significantly outperforms training the same network on either random unclicked news or on the dataset's own impression logs. The paper reads this as evidence that data quality, not architectural complexity, is the limiting factor, and that the same negative-sampling idea is what makes the decentralized deployment feasible.

Load-bearing premise

The load-bearing premise is that a user's least similar news articles, measured by cosine distance from the average of their reading history, are genuinely uninteresting to that user, so labeling them as negatives is safe.

Editorial extensions

If this is right

  • On the MIND dataset, the same tiny network that performs near randomly when trained on impression negatives reaches state-of-the-art-level AUC when trained on synthetic farthest-item negatives.
  • The accuracy gain is statistically significant over both random negative sampling and impression-based negatives, so the choice of negative labels is a first-order design decision.
  • Training a separate lightweight model per user is computationally practical, with bulk prediction for 100,000 users estimated at roughly 88 minutes at the 60-sample cap.
  • Because the synthetic pool for a user depends only on past activity, it can be precomputed offline, and only the final scoring needs to run near real time.
  • The same content-only training pipeline can run on user devices or edge nodes, since user history never has to leave the device and only item embeddings are computed centrally.

Reading between the lines

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

  • The paper does not claim this, but the same farthest-item rule should transfer to other content domains with dense embeddings, such as video or product recommendation.
  • A stress test the paper leaves open: restrict the negative pool to articles in the same category as the clicked history; if the AUC advantage shrinks, the method is partly exploiting easy category boundaries rather than fine-grained taste.
  • The paper does not describe a time-based split, so an important check is whether drawing negatives from the full news pool, including articles that appear later in test impressions, inflates the reported AUC.
  • A live deployment with explicit or behavioral disinterest signals would tell whether farthest-item negatives track real user preferences better than random sampling.
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 / 6 minor

Summary. The paper proposes DNNR, a content-based news recommender that trains one small feed-forward network per user. The central methodological contribution is a negative sampling strategy (Algorithm 1) that builds a 'Synthetic Pool' for each user by taking the farthest news embeddings from the user's averaged clicked-history embedding in an L2-normalized space, using a multilingual MiniLM encoder. The authors report that this sampling strategy outperforms training on impressions or on random negatives (Section 4.3), that the per-user networks achieve AUC comparable to NRMS, DKN, and NPA on MIND (Section 4.4, Figure 15), and that the approach lends itself to decentralized deployment (Section 5.2). The paper also studies the effect of the maximum sample size on AUC and training/prediction times (Sections 4.2 and 4.5).

Significance. If the central claim is validated, the paper would make a useful practical contribution: a lightweight, per-user content-based model whose performance comes primarily from a carefully constructed negative sample rather than from a deep architecture. The internal ablation in Section 4.3 is a genuine strength: the comparisons between Synthetic Pool, Random Sampling, and Impressions are subjected to statistical tests (Welch ANOVA, Games-Howell post-hoc, and a per-user Kruskal-Wallis analysis), and the writing is candid about limitations. The decentralization discussion is coherent and the limitations section explicitly acknowledges that the approach 'cannot beat' more complex architectures. However, the headline comparison with the state of the art is not currently supported: the evaluation protocol is under-specified with respect to temporal splits, and Figure 15 reports no exact values, error bars, or significance tests. These issues are load-bearing for the abstract's claim that the method 'can compete with that of State-of-the-Art models.'

major comments (4)
  1. [Section 3.5, Algorithm 1; Section 4.4, Figure 15] Temporal leakage in the negative sampler is a serious threat to the reported AUC values. Algorithm 1 forms the Synthetic Pool for user u by sorting all news embeddings by inner product with the user centroid and taking the farthest Sample Length items. The search space is described as the full embedding index, with no restriction to news items published before the user's impression timestamp, and no exclusion of items that appear in the user's test impressions. Since MIND contains click histories and impressions at different timestamps for the same user, an article that appears in a user's test impressions can be selected, before evaluation, as one of that user's synthetic negatives. The model is then explicitly trained to assign low probability to the exact embedding that is later scored at test time. This can inflate the DNNR AUC in Section 4.3 and explain part of the parity with NRMS/DKN/NPA in Figure 15. The authors must rerun the experiments after restricting the negative-sampling pool to news items whose timestamps precede the user's impression time and excluding all impression items; if the official MIND train/dev/test split is used, that split and the exact protocol must be stated.
  2. [Section 4.4, Figure 15] The state-of-the-art comparison is underreported. Figure 15 shows only mean values for AUC, without error bars, confidence intervals, the number of users or impressions used, or any significance test comparing DNNR to NRMS, DKN, or NPA. The manuscript also does not report the exact AUC values or the standard deviations. Because the central claim of the abstract rests on this comparison, the authors should provide this information, ideally as per-user paired AUC comparisons with a suitable statistical test, and should describe how the baseline scores were obtained (e.g., from the original papers or from local re-runs) and which MIND split was used.
  3. [Section 4.3, Figures 12 and 13] The internal ablation that supports the proposed negative sampling mechanism is also affected by the missing temporal split. The Synthetic Pool, Random Sampling, and Impressions training conditions all draw negative examples from a pool that may contain articles appearing in the test impressions; at minimum, the paper should confirm that none of the three conditions uses test-candidate embeddings during training, or rerun the comparison with a proper temporal split. Until this is done, the clean separation in the PCA plots (Figure 14) and the AUC differences may partly reflect different degrees of test-candidate contamination rather than purely the quality of the negative sampling strategy.
  4. [Section 4.2, Figures 10 and 11] The choice of the maximum sample size hyperparameter (60) is made by comparing AUC values computed on the impressions that are later used for the final evaluation. This is model selection on the test set, which overstates the expected performance of the selected configuration. The paper should either use a validation split for the hyperparameter choice or explicitly acknowledge that the reported AUC values are optimistic because the hyperparameter was selected on the evaluation data.
minor comments (6)
  1. [Section 1] There is a typo in the contributions list: 'Summarizing, we our contributions are the following' should read 'Summarizing, our contributions are the following.'
  2. [Section 3.3] 'forth layer' should be 'fourth layer'.
  3. [Section 3.5, Algorithm 1] The pseudocode is ambiguous about the sort order: to select the farthest items, the inner products should be sorted in increasing order, but the line 'inner product indices← Sort(⟨centroid, embeddings⟩)[indices]' does not specify ascending or descending. The text should state this explicitly for reproducibility.
  4. [Section 4.3] The test is referred to as the 'Welsh test'; the standard spelling is 'Welch test'.
  5. [Section 4.5] There are several typos: 'maitaining' should be 'maintaining', 'the architecture of the hole system' should be 'the architecture of the whole system', and 'it would be need around 88 minutes' should be 'it would need around 88 minutes'.
  6. [Section 4.4] The citation [31] (Bradley, 1997) is a general reference for AUC and does not support the specific baseline values shown in Figure 15; the source of the NRMS/DKN/NPA numbers should be identified explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central claim is evaluated against external baselines on an external dataset, and the negative-sampling construction does not reduce to the reported prediction.

full rationale

The paper's central claim is that a farthest-item negative sampling strategy improves a lightweight per-user network to compete with NRMS, DKN, and NPA on the MIND dataset. The derivation chain is not circular: the negative sampler constructs training labels from embedding distances to a user centroid (Algorithm 1, Section 3.5), and a separate feed-forward network is trained on those labels and evaluated on held-out impression logs. The test labels are real clicks and non-clicks from MIND impressions, not the synthetic labels, so the reported AUC is not identical by construction to the sampling rule. The comparison is made against external baselines (NRMS, DKN, NPA) whose results are cited from prior work and not derived from the paper's own fitted values. The only self-citation, reference [20] on recency-based negative feedback, appears in related work and is not load-bearing; no uniqueness theorem or prior author result is invoked to force the method choice. Selecting the 60-sample cap using AUC on the evaluation data is test-set tuning and a validity concern, but it is not a circularity in the sense of a prediction reducing to its input. Potential temporal leakage in Algorithm 1 is a serious experimental validity risk but not a definitional equivalence, so it does not affect the circularity score.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The method introduces no new physical or conceptual entity; it relies on a domain assumption about embedding distance and on the standard L2 identity. The free parameters (sample cap, epochs, architecture) were selected using the same evaluation data, which is a correctness risk rather than a circularity.

free parameters (3)
  • max_samples = 60
    Chosen in Section 4.2 as the best trade-off between group AUC and computation time (Kruskal-Wallis and ANOVA on the evaluation split). This hyperparameter is part of the reported system.
  • epochs = 15
    Set because loss values stabilize around 15 iterations (Section 4, Figure 5), with no independent validation split described.
  • architecture_size = 4 layers reducing 384 to 64 dims, dropout 0.2
    Network layers and dropout are selected without systematic search; they influence the accuracy comparison in Figure 15.
assumptions (5)
  • standard math For L2-normalized vectors, squared Euclidean distance is proportional to cosine distance (Equation 1).
    Used to convert maximum inner product search into farthest-item selection in Algorithm 1; this is a standard vector identity.
  • domain assumption A user's interest can be represented as the average embedding (centroid) of the titles of the news they read.
    Algorithm 1 computes the user centroid as the averaged embeddings of all titles from the user's history; no evidence is given that a centroid is a good user model.
  • domain assumption News items farthest from the user centroid in the embedding space are valid negative examples.
    The synthetic negative pool is built from the lowest inner-product items; the paper argues these are 'putatively uninteresting news' (Section 4.3) but does not validate against explicit user feedback.
  • domain assumption The evaluation protocol, using AUC on impression logs, measures real recommendation quality.
    The paper uses group and individual AUC on MIND impressions (Section 4.4). It acknowledges that impression items are already filtered by a recommender and may be relevant to the user, which is a known limitation.
  • domain assumption The entire MIND news collection can be searched when constructing synthetic negatives without regard to the timestamp of the user's impressions.
    Algorithm 1 sorts over all embeddings; the paper does not restrict the search to news published before the impression time, so temporal leakage is possible.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rethinking negative sampling in content-based news recommendation." pith.science (2026). https://pith.science/paper/CHKO7KZC

@misc{pith2026241108700,
  author       = {Pith},
  title        = {Pith review of: Rethinking negative sampling in content-based news recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CHKO7KZC}},
  note         = {Machine review of arXiv:2411.08700}
}
read the original abstract

News recommender systems are hindered by the brief lifespan of articles, as they undergo rapid relevance decay. Recent studies have demonstrated the potential of content-based neural techniques in tackling this problem. However, these models often involve complex neural architectures and often lack consideration for negative examples. In this study, we posit that the careful sampling of negative examples has a big impact on the model's outcome. We devise a negative sampling technique that not only improves the accuracy of the model but also facilitates the decentralization of the recommendation system. The experimental results obtained using the MIND dataset demonstrate that the accuracy of the method under consideration can compete with that of State-of-the-Art models. The utilization of the sampling technique is essential in reducing model complexity and accelerating the training process, while maintaining a high level of accuracy. Finally, we discuss how decentralized models can help improve privacy and scalability.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 18 canonical work pages

  1. [1]

    Slifka, M.K., Whitton, J.L.: Clinical implications of dysregulated cytokine production. J. Mol. Med. 78, 74–80 (2000). https://doi.org/10.1007/ 26 Rethinking negative sampling in content-based news recommendation s001090000086

    work page 2000

  2. [2]

    Information Processing & Management 54(6), 1203–1227 (2018)

    Karimi, M., Jannach, D., Jugovac, M.: News recommender systems – Survey and roads ahead. Information Processing & Management 54(6), 1203–1227 (2018). https://doi.org/10.1016/j.ipm.2018.04.008

    doi:10.1016/j.ipm.2018.04.008 2018

  3. [3]

    IEEE Communications Surveys & Tutorials 22(2), 869–904 (2020)

    Wang, X., Han, Y., Leung, V.C.M., Niyato, D., Yan, X., Chen, X.: Convergence of Edge Computing and Deep Learning: A Comprehen- sive Survey. IEEE Communications Surveys & Tutorials 22(2), 869–904 (2020). https://doi.org/10.1109/COMST.2020.2970550

    arXiv 2020

  4. [4]

    In: van der Aalst, W., Mylopoulos, J., Rosemann, M., Shaw, M.J., Szyper- ski, C., Huemer, C., Lops, P

    Jannach, D., Zanker, M., Ge, M., Gr¨ oning, M.: Recommender Systems in Computer Science and Information Systems – A Landscape of Research. In: van der Aalst, W., Mylopoulos, J., Rosemann, M., Shaw, M.J., Szyper- ski, C., Huemer, C., Lops, P. (eds.) E-Commerce and Web Technologies vol. 123, pp. 76–87. Springer Berlin Heidelberg, Berlin, Heidelberg (2012). ...

    work page doi:10.1007/978-3-642-32273-0 2012

  5. [5]

    In: Proceedings of the 1st International Work- shop on Data Semantics - DataSem ’10, p

    IJntema, W., Goossen, F., Frasincar, F., Hogenboom, F.: Ontology-based news recommendation. In: Proceedings of the 1st International Work- shop on Data Semantics - DataSem ’10, p. 1. ACM Press, Lausanne, Switzerland (2010). https://doi.org/10.1145/1754239.1754257

    work page arXiv 2010

  6. [6]

    Rao, J., Jia, A., Feng, Y., Zhao, D.: Personalized News Recommendation Using Ontologies Harvested from the Web. In: Hutchison, D., Kanade, T., Kittler, J., Kleinberg, J.M., Mattern, F., Mitchell, J.C., Naor, M., Nierstrasz, O., Pandu Rangan, C., Steffen, B., Sudan, M., Terzopoulos, D., Tygar, D., Vardi, M.Y., Weikum, G., Wang, J., Xiong, H., Ishikawa, Y.,...

    work page 2013

  7. [7]

    Expert Systems with Applications 32(1), 143–150 (2007)

    Lee, H.J., Park, S.J.: MONERS: A news recommender for the mobile web. Expert Systems with Applications 32(1), 143–150 (2007). https:// doi.org/10.1016/j.eswa.2005.11.010

    work page doi:10.1016/j.eswa.2005.11.010 2007

  8. [8]

    WWW ’07, pp

    Ahn, J.-w., Brusilovsky, P., Grady, J., He, D., Syn, S.Y.: Open user profiles for adaptive news systems: Help or harm? In: Proceedings of the 16th International Conference on World Wide Web. WWW ’07, pp. 11–20. Association for Computing Machinery, New York, NY, USA (2007). https: //doi.org/10.1145/1242572.1242575

    arXiv 2007

Show all 36 references
  1. [9]

    In: Proceedings of the 16th International Conference on World Wide Web

    Das, A.S., Datar, M., Garg, A., Rajaram, S.: Google news personaliza- tion: Scalable online collaborative filtering. In: Proceedings of the 16th International Conference on World Wide Web. WWW ’07, pp. 271–

  2. [10]

    In: Proceedings of the 15th International Con- ference on Intelligent User Interfaces - IUI ’10, p

    Liu, J., Dolan, P., Pedersen, E.R.: Personalized news recommendation based on click behavior. In: Proceedings of the 15th International Con- ference on Intelligent User Interfaces - IUI ’10, p. 31. ACM Press, Hong Kong, China (2010). https://doi.org/10.1145/1719970.1719976

    2010

  3. [11]

    International Journal of Computer and Com- munication Engineering 5(4), 276–285 (2015)

    Vietnam National University — University of Engineering and Technol- ogy, Hanoi, Vietnam., Huy, N.T., Chuan, D.T., Nguyen, V.A.: Implicit Feedback Mechanism to Manage User Profile Applied in Vietnamese News Recommender System. International Journal of Computer and Com- municat...

    2015

  4. [12]

    In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp

    Wu, C., Wu, F., An, M., Huang, J., Huang, Y., Xie, X.: NPA: Neural News Recommendation with Personalized Attention. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2576–2584. ACM, Anchorage AK USA (2019). https://doi. o...

    2019

  5. [13]

    In: Proceedings of the 31st International Conference on Neural Information Processing Systems

    Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, L., Polosukhin, I.: Attention is all you need. In: Proceedings of the 31st International Conference on Neural Information Processing Systems. NIPS’17, pp. 6000–6010. Curran Associates Inc., Re...

    2017

  6. [14]

    In: Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW ’18, pp

    Wang, H., Zhang, F., Xie, X., Guo, M.: DKN: Deep Knowledge-Aware Network for News Recommendation. In: Proceedings of the 2018 World Wide Web Conference on World Wide Web - WWW ’18, pp. 1835–

    2018

  7. [15]

    Wu, C., Wu, F., Ge, S., Qi, T., Huang, Y., Xie, X.: Neural News Rec- ommendation with Multi-Head Self-Attention. In: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Process- ing and the 9th International Joint Conference on Natural Language Processi...

    2019

  8. [16]

    arXiv:2104.07413 [cs] (2021) https: //arxiv.org/abs/2104.07413 [cs]

    Wu, C., Wu, F., Qi, T., Huang, Y.: Empowering News Recommendation with Pre-trained Language Models. arXiv:2104.07413 [cs] (2021) https: //arxiv.org/abs/2104.07413 [cs]

    2021 arXiv

  9. [17]

    CCF Transactions on Pervasive Com- puting and Interaction 2(3), 178–188 (2020)

    Wu, C., Wu, F., Huang, Y., Xie, X.: Neural news recommenda- tion with negative feedback. CCF Transactions on Pervasive Com- puting and Interaction 2(3), 178–188 (2020). https://doi.org/10.1007/ s42486-020-00044-0 28 Rethinking negative sampling in content-based news recommendation

    2020

  10. [18]

    In: 2019 6th Inter- national Conference on Systems and Informatics (ICSAI), pp

    Feng, W.-B., Zhang, Z.-S., Li, C.-Z., Song, A.-P.: Negative examples sampling based on factorization machines for OCCF. In: 2019 6th Inter- national Conference on Systems and Informatics (ICSAI), pp. 1482–1485 (2019). https://doi.org/10.1109/ICSAI48974.2019.9010476

    2019

  11. [19]

    In: Ohsawa, Y., Yada, K., Ito, T., Takama, Y., Sato-Shimokawara, E., Abe, A., Mori, J., Matsumura, N

    Zhang, J.-c., Takama, Y.: Proposal of context-aware music recommender system using negative sampling. In: Ohsawa, Y., Yada, K., Ito, T., Takama, Y., Sato-Shimokawara, E., Abe, A., Mori, J., Matsumura, N. (eds.) Advances in Artificial Intelligence, pp. 114–125. Springer Interna...

    2020

  12. [20]

    In: Proceedings of the 30th Annual ACM Symposium on Applied Computing

    Vinagre, J.a., Jorge, A.M., Gama, J.a.: Collaborative filtering with recency-based negative feedback. In: Proceedings of the 30th Annual ACM Symposium on Applied Computing. SAC ’15, pp. 963–965. Association for Computing Machinery, New York, NY, USA (2015). https://doi.org/ 10...

    2015

  13. [21]

    In: Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp

    Okura, S., Tagami, Y., Ono, S., Tajima, A.: Embedding-based News Rec- ommendation for Millions of Users. In: Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1933–1942. ACM, Halifax NS Canada (2017). https://doi. org/10.11...

    2017

  14. [22]

    In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp

    Wang, H., Wu, F., Liu, Z., Xie, X.: Fine-grained interest matching for neu- ral news recommendation. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 836–845. Association for Computational Linguistics, Online (2020). https://doi....

    2020

  15. [23]

    Science China Technologi- cal Sciences 63(10), 1872–1897 (2020) https://arxiv.org/abs/2003.08271

    Qiu, X., Sun, T., Xu, Y., Shao, Y., Dai, N., Huang, X.: Pre-trained Models for Natural Language Processing: A Survey. Science China Technologi- cal Sciences 63(10), 1872–1897 (2020) https://arxiv.org/abs/2003.08271. https://doi.org/10.1007/s11431-020-1647-3

    2020 arXiv

  16. [24]

    In: Pro- ceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval

    Canny, J.: Collaborative filtering with privacy via factor analysis. In: Pro- ceedings of the 25th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval. SIGIR ’02, pp. 238–

  17. [25]

    Expert Systems with Applications 27(2), 203–210 (2004)

    Han, P., Xie, B., Yang, F., Shen, R.: A scalable p2p recommender sys- tem based on distributed collaborative filtering. Expert Systems with Applications 27(2), 203–210 (2004). https://doi.org/10.1016/j.eswa.2004. 01.003

    2004 doi

  18. [26]

    ACM Trans

    Miller, B.N., Konstan, J.A., Riedl, J.: Pocketlens: Toward a personal Rethinking negative sampling in content-based news recommendation 29 recommender system. ACM Trans. Inf. Syst. 22(3), 437–476 (2004). https://doi.org/10.1145/1010614.1010618

    2004

  19. [27]

    In: Proceedings of the 13th ACM Conference on Recommender Systems

    Eichinger, T., Beierle, F., Papke, R., Rebscher, L., Tran, H.C., Trzeciak, M.: On gossip-based information dissemination in pervasive recommender systems. In: Proceedings of the 13th ACM Conference on Recommender Systems. RecSys ’19, pp. 442–446. Association for Computing Mach...

    2019

  20. [28]

    ArXiv abs/1207.0580 (2012)

    Hinton, G.E., Srivastava, N., Krizhevsky, A., Sutskever, I., Salakhutdinov, R.: Improving neural networks by preventing co-adaptation of feature detectors. ArXiv abs/1207.0580 (2012)

    2012 arXiv

  21. [29]

    Wang, W., Wei, F., Dong, L., Yang, N., Zhou, M.: MINILM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers, 13 (2020)

    2020

  22. [30]

    In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp

    Wu, F., Qiao, Y., Chen, J.-H., Wu, C., Qi, T., Lian, J., Liu, D., Xie, X., Gao, J., Wu, W., Zhou, M.: MIND: A Large-scale Dataset for News Recommendation. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 3597–3606. Association fo...

    2020 doi

  23. [31]

    Pattern Recognition 30, 1145–1159 (1997)

    Bradley, A.P.: The use of the area under the ROC curve in the evalua- tion of machine learning algorithms. Pattern Recognition 30, 1145–1159 (1997)

    1997

  24. [32]

    Devlin, J., Chang, M.-W., Lee, K., Toutanova, K.: BERT: Pre-training of deep bidirectional transformers for language understanding. In: Pro- ceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technolo-...

    2019 doi

  25. [33]

    Reimers, N., Gurevych, I.: Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. In: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th Inter- national Joint Conference on Natural Language Processing (EMNLP- IJCNLP), pp...

    2019 doi

  26. [245]

    https://doi.org/10.1145/564376.564419

    Association for Computing Machinery, New York, NY, USA (2002). https://doi.org/10.1145/564376.564419. https://doi.org/10.1145/564376. 564419

    2002

  27. [280]

    https://doi.org/10.1145/1242572.1242610 Rethinking negative sampling in content-based news recommendation 27

    Association for Computing Machinery, New York, NY, USA (2007). https://doi.org/10.1145/1242572.1242610 Rethinking negative sampling in content-based news recommendation 27

    2007

  28. [1844]

    https://doi.org/10.1145/3178876

    ACM Press, Lyon, France (2018). https://doi.org/10.1145/3178876. 3186175

    2018 doi

Pith tools

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