An End-to-End Neighborhood-based Interaction Model for Knowledge-enhanced Recommendation

Jian Tang; Jianyun Nie; Ting Bai; Weinan Zhang; Yanru Qu

REVIEW 3 major objections 8 minor 37 references

An End-to-End Neighborhood-based Interaction Model for Knowledge-enhanced Recommendation

T0 review · 3 major / 8 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A bi-attention over neighbor pairs improves graph-based recommendation, reporting 1.1–8.4% AUC gains over strong baselines on four datasets.

desk verdict A genuinely new interaction mechanism with a clean unifying view, but the headline AUC gains rest on a sampled-negative protocol that needs more scrutiny before I would trust the absolute numbers. read the letter →

arxiv 1908.04032 v2 pith:7GDRYHT3 submitted 2019-08-12 cs.IR

Yanru Qu , Ting Bai , Weinan Zhang , Jianyun Nie , Jian Tang This is my paper · ORCID

classification cs.IR

keywords KnowledgeGraphKnowledge-enhancedRecommendationNeighborhood-basedInteractionneuralnetworkclick-throughratepredictiontop-Nbi-attentionearlysummarization

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 identifies an 'early summarization' flaw in graph-based recommender systems: they compress each user's and item's neighborhoods into a single vector before scoring, flattening many useful neighbor-pair signals into one embedding. It proposes the Neighborhood Interaction model, which scores every pair of user-side and item-side neighbors with a learned bi-attention weight, and extends it with graph neural networks and external knowledge graphs to form KNI. The paper reports that KNI improves click-through rate prediction by 1.1%–8.4% in absolute AUC over strong feature-based, meta-path, and graph baselines across four real-world datasets, and wins by a wide margin in top-N recommendation. If correct, the message is that graph-based recommenders should preserve pairwise neighborhood structure instead of summarizing early, with the largest gains on sparse data.

What carries the argument

The central object is the neighborhood-interaction decomposition $\hat{y} = A \odot Z$, read as a weighted sum over all pairs $(i,j)$ of user-side neighbor $i$ and item-side neighbor $j$: $A_{i,j}$ is a nonnegative weight summing to 1 and $Z_{i,j} = \langle x_i, x_j \rangle$. The paper shows that average aggregation and attention aggregation are special cases of this form, with $A$ constant or rank-one; NI instead learns $A$ directly through a bi-attention network $\alpha_{i,j} = \mathrm{softmax}(w^\top[x_u, x_i, x_v, x_j] + b)$. This machinery is what lets the model give distinct weight to informative neighbor pairs, such as a liked romance film matching a romance tag, while down-weighting noise, such as a liked film matching an unrelated tag.

What would settle it

Rerun KNI and the attention-aggregation baseline on data where explicit low ratings (1–2 stars) serve as negatives instead of sampled unseen items; if KNI's AUC and top-N advantages shrink or disappear, the reported superiority depends on the unobserved-as-negative assumption rather than on neighborhood interaction.

Watch

Extended reading notes

Core claim

The paper's central claim is that standard graph-based recommendation pipelines throw away information by aggregating neighborhoods into two vectors before computing a prediction. The proposed fix is to score the neighborhoods directly: for a user $u$ and item $v$, $\hat{y}_{u,v}$ is a weighted sum over all pairs $(i,j)$ where $i$ is a user-side neighbor and $j$ is an item-side neighbor, with weights assigned by a bi-attention network that sees both neighborhoods at once. The paper shows this re-expresses prior graph models in the general form $\hat{y} = A \odot Z$, with the average aggregator as a constant weight matrix and the attention aggregator as a rank-one weight matrix; NI replaces both with a fully learned weight matrix. Adding graph convolution and attention encoders and knowledge-graph entities yields KNI. The paper's reported result is that KNI beats the strongest compared methods by 1.1%–8.4% absolute AUC in CTR prediction and by a wide margin in top-N recommendation on four real-world datasets, with the biggest gains on the two sparser book datasets.

Load-bearing premise

The argument's load-bearing premise is that unseen items are negatives: for each user the paper samples the same number of '0' labels from unseen items and computes AUC and top-N metrics against that sample, so if many unobserved items would actually be liked, the reported gains may not reflect true ranking quality.

Editorial extensions

If this is right

NI outperforms PinSage and RippleNet even without the knowledge graph, so pairwise neighborhood interaction itself, not the KG, drives much of the gain.
The absolute AUC advantage over the best baseline is largest on the sparsest datasets (4.33% on C-Book, 5.02% on A-Book), so the method helps most where interactions are scarce.
High-order neighbors are useful only up to a point: performance drops at 3–4 hops because neighborhood sizes explode and add noise, and one hop is enough for KNI.
With neighbor sampling, KNI trains 5.6–8.6 times faster per iteration than RippleNet in the reported settings, so the added expressiveness does not require slower training.

Reading between the lines

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

Beyond the paper's experiments, the $\hat{y} = A \odot Z$ form suggests a diagnostic: when a model's attention-weight matrix has near-zero entropy, it has collapsed to one neighbor pair, so such cases should be over-represented among errors of attention-aggregation models.
A direct ablation test follows from the paper's decomposition: hold the GNN encoder and KG fixed and replace the bi-attention with the rank-one attention form; the gap between the two versions would quantify how much of KNI's gain comes specifically from joint neighbor weighting.
The paper treats all unobserved pairs as negatives in training; an untested extension is to vary the negative-sampling ratio or use explicit low ratings as negatives, which would separate the model's ranking ability from the evaluation assumption.
The decomposition also hints at a practical scaling route the paper does not pursue: for very large neighborhoods, sparsifying low-weight pairs in $A$ before the weighted sum could preserve most of the signal while cutting computation.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague, quick take on 1908.04032. The real contribution is the Neighborhood Interaction idea: instead of summarizing user and item neighborhoods into single embeddings and taking an inner product, it scores each (user-neighbor, item-neighbor) pair with a bi-attention weight. The Eq. (11) reformulation showing that average and attention aggregation are special cases is clean and genuinely useful. The experimental scope is also broader than most: four datasets, eight baselines, code and data released, and a training-time comparison showing KNI is 5-8x faster than RippleNet. The model definition is coherent, and I see no circularity in the core claim.

The soft spot is exactly what the stress-test note flags. The evaluation samples, for each user, as many negatives from unseen items as the user has positives. All reported AUC/ACC and top-N numbers are computed against that constructed set. If the sampled unseen items are not representative of what the user would actually skip, the 1.1-8.4% absolute gains are inflated. The paper does not analyze the negative distribution and reports no standard deviations. Section 3.3 says experiments are repeated 5 times, then later says 40 evaluations were used; the p<10^-6 claim is not auditable without variance. The hop-number choice is made on validation, which is standard, and the entropy case study is indirect evidence for the mechanism, not a direct test.

Is this fatal? No. The protocol is common in one-class collaborative filtering, and all baselines face the same constructed negatives, so the relative ordering may survive. But the abstract's absolute improvements should be read with caution until the negative-sampling bias is characterized. The exclusion of LFM from KB4Rec is thinly justified, but that is minor.

This paper deserves a serious referee. With error bars, a negative-sampling robustness check, and clearer repetition reporting, it could be a solid journal paper. For now, cite it for the NI concept and the unifying form, not for the empirical magnitude.

Referee Report

3 major / 8 minor

Summary. The paper addresses graph-based recommendation and identifies an 'early summarization' problem: existing models aggregate user- and item-side neighborhoods into single embeddings before scoring, losing pairwise structural information. It proposes Neighborhood Interaction (NI), which scores a user-item pair by a bi-attention-weighted sum of inner products over all user-neighbor/item-neighbor pairs, and Knowledge-enhanced NI (KNI), which augments the interaction graph with a knowledge graph and applies GNN propagation. The central claim, stated in the abstract and Section 3.4, is that KNI outperforms state-of-the-art feature-based, meta path-based, and KG-based recommenders by 1.1%-8.4% absolute AUC and by a wide margin in top-N recommendation on four real-world datasets. The paper also provides a case study aiming to show that NI learns more informative attention distributions than attention aggregation models.

Significance. If the empirical claims hold, KNI is a meaningful contribution: it offers a clean reformulation of neighborhood-based scoring (Eq. (11)) that unifies average and attention aggregation as special cases, and it gives a concrete, end-to-end architecture with publicly available code and datasets. The derivation of the pairwise interaction form and the bi-attention model is transparent and reproducible. The claimed gains on sparse book datasets are particularly interesting, as they suggest the model handles sparsity well. However, the current evaluation does not fully establish the magnitude of the advantage, because the reported numbers depend on a sampled-negative protocol and lack variance information.

major comments (3)

[Section 3.1, Table 2, Figs. 3-6] The central empirical claim rests on a negative-sampling protocol that is not fully specified. For A-Book and Movie-20M, Section 3.1 says 'For each user, we sample the same amount of negative samples ... from unseen items'; it is not stated how many negatives are sampled at test time, whether the same negative set is used for all models, or whether any negative sampling is applied to C-Book and Movie-1M. AUC and top-N metrics computed against a small or easy negative set can inflate absolute gains, because KNI's neighborhood representations may separate sampled negatives more easily than baselines. The authors should report the negative-set construction in detail, evaluate on a fixed large negative set or full item ranking for test users, and show sensitivity to the negative sampling ratio. This is required to support the 1.1%-8.4% absolute AUC claim.
[Section 3.3 and Table 2] The paper reports no standard deviations or confidence intervals, despite saying 'We repeat each experiment 5 times and report the average scores' and using 40 evaluations per run (Fig. 2). The p<1e-6 significance asterisks in Table 2 cannot be audited without knowing the unit of analysis (5 runs? 40 evaluations?) and the variance across runs. This matters because the smallest claimed improvements are around 1% absolute AUC (Table 3, Movie-20M: 1.07%), which may be within run-to-run variability. Please report mean ± std or a confidence interval over repeated runs for all metrics.
[Section 3.6 and Fig. 7] The case study's conclusion that NI 'has the capability to learn more informative neighborhood interactions' is based on entropy histograms from a single random subset and one hand-picked test pair. No statistical test accompanies the average entropy values (2.12 vs. 3.18), and higher attention entropy is not established as a reliable proxy for predictive quality. This evidence is suggestive but does not by itself confirm the early-summarization mechanism. Please either add significance testing and variance over multiple subsets, or soften the causal interpretation.

minor comments (8)

[Section 2.1.3] There is a typo: 'asnd' should be 'and' in the sentence describing the average form as a special case.
[Section 2.2] The phrase 'we can recklessly replace Gr ec with G' should likely be 'readily replace' or 'directly replace'; 'recklessly' is not the intended meaning.
[Section 3.6] The sentence 'we conduct a case study on in this section' is grammatically incomplete and should be revised.
[Section 3 and Table 4] The paper does not state which GNN encoder (GCN or GAT) is used in the final KNI model for the experiments; Section 2.2 presents both, but Section 3 never specifies which architecture produced Tables 2-4. This should be clarified for reproducibility.
[Section 3.1 and Table 3] It is unclear whether C-Book and Movie-1M use the same negative-sampling protocol as A-Book and Movie-20M, and whether the n-hop columns in Table 3 are average numbers of neighbors per node or total counts; please clarify both points.
[Section 3.3 and Fig. 2] The relationship between 'repeat each experiment 5 times' and 'fix this number to 40' is ambiguous; please state explicitly how the 40 evaluations and 5 repeats are combined to produce the reported averages and significance tests.
[Section 2.2, Eq. (19)] The notation for attention parameters (wla, bla) is inconsistent with the GCN notation (w, b) used earlier; please unify the notation.
[Figure 8] The color bars have different scales across panels (a)-(d), making visual comparison of AAM and NI difficult; consider using a shared color scale.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: KNI's reported gains come from held-out evaluation, and Eq. (11) is a reformulation, not a fitted input.

full rationale

The paper's central claim is empirical: KNI outperforms eight baselines on CTR prediction and top-N recommendation. That claim is supported by training with the log-loss of Eq. (21), validation-based hyperparameter selection, and evaluation on held-out test splits. The mathematical core, Eq. (11), is explicitly derived as the common algebraic form of the average and attention aggregation schemes in Eqs. (8) and (10), so it is a reformulation of existing model forms rather than a fitted quantity or an assumed conclusion. The bi-attention NI model in Eqs. (13)-(14) is a new architectural choice and is not defined in terms of the reported AUC improvements. GCN/GAT components are standard external architectures cited from [15,24], and the knowledge-graph construction follows RippleNet/KB4Rec, which are not authored by this paper's authors in a way that would make the comparison circular. The only self-references, to the authors' earlier PNN papers in the related-work enumeration, are not load-bearing for the central claim. The sampled-negative evaluation protocol is a legitimate concern about external validity, but it is not circularity: the test negatives are not used to fit the model, and the reported predictions are not a re-statement of the sampled labels. No derivation step reduces by construction to its own input.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central experimental claim rests on a set of domain assumptions about how feedback data are interpreted and how the knowledge graph is integrated. The most important is the binary treatment of ratings and the use of sampled unseen items as negatives, which is a standard but strong assumption in implicit feedback evaluation. The accuracy of the item-entity mapping from KB4Rec is also assumed. In addition, the neighbor sampling strategy is assumed to introduce negligible bias. These assumptions are reasonable for the recommendation literature but should be kept in mind when interpreting the reported gains.

free parameters (5)

hop_number = 1 (KNI), 1 (PinSage), 3/2/1/1 (RippleNet per dataset)
The number of graph propagation hops is tuned on validation AUC. KNI and PinSage perform best with 1 hop, meaning high-order neighbors add noise. This choice directly affects the structure of neighbors used in the model.
neighbor_sample_size = 4 (C-Book), 32 (Movie-1M), 8 (A-Book), 32 (Movie-20M)
The number of sampled neighbors K is tuned from {4,8,16,32,64,128}. It controls the computational cost and the neighborhood coverage; the reported values are selected by validation AUC.
embedding_dimension = 128
Grid search over {4,8,16,32,64,128}; 128 gives the best validation AUC. Higher dimensions were not tested due to memory limits.
learning_rate = 0.001
Grid search over {1e-4, 2e-4, 5e-4, 1e-3, 2e-3, 5e-3, 1e-2}; 1e-3 is generally best across models.
l2_regularization = 1e-5 (C-Book), 1e-7 (Movie-1M), 1e-7 (A-Book), 1e-8 (Movie-20M)
Set differently per dataset based on validation AUC; this regularizer controls overfitting and is part of the training objective in Eq. (21).

assumptions (4)

domain assumption Unseen items are treated as negative examples for training and evaluation.
Section 3.1: for each user, an equal number of negative samples are drawn from unseen items. This assumes non-interaction implies dislike, which is not guaranteed in implicit feedback.
domain assumption The knowledge graph linkage from items to entities is accurate for A-Book and Movie-20M.
The paper relies on KB4Rec [32] for item-entity mapping; noisy or incomplete mapping would add noise to the knowledge-enhanced interaction graph.
domain assumption Neighbor sampling (NS) with fixed size K preserves the information needed for the model.
Section 2.2 adopts NS to limit neighborhood size; this assumes sampled neighbors are representative of the full neighborhood.
domain assumption The binary conversion of ratings (4/5 positive, others negative) preserves the signal for CTR prediction.
Section 3.1: ratings are binarized; this discards rating intensity, which may matter for some recommendation scenarios.

how reviews work

0 comments

Cite this review

Pith. "Pith review of An End-to-End Neighborhood-based Interaction Model for Knowledge-enhanced Recommendation." pith.science (2026). https://pith.science/paper/7GDRYHT3

@misc{pith2026190804032,
  author       = {Pith},
  title        = {Pith review of: An End-to-End Neighborhood-based Interaction Model for Knowledge-enhanced Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7GDRYHT3}},
  note         = {Machine review of arXiv:1908.04032}
}

read the original abstract

This paper studies graph-based recommendation, where an interaction graph is constructed from historical records and is lever-aged to alleviate data sparsity and cold start problems. We reveal an early summarization problem in existing graph-based models, and propose Neighborhood Interaction (NI) model to capture each neighbor pair (between user-side and item-side) distinctively. NI model is more expressive and can capture more complicated structural patterns behind user-item interactions. To further enrich node connectivity and utilize high-order structural information, we incorporate extra knowledge graphs (KGs) and adopt graph neural networks (GNNs) in NI, called Knowledge-enhanced NeighborhoodInteraction (KNI). Compared with the state-of-the-art recommendation methods,e.g., feature-based, meta path-based, and KG-based models, our KNI achieves superior performance in click-through rate prediction (1.1%-8.4% absolute AUC improvements) and out-performs by a wide margin in top-N recommendation on 4 real-world datasets.

Figures

Figures reproduced from arXiv: 1908.04032 by the authors.

**Figure 1.** Model overview. Note: Red circles denote users. Green circles denote rated or unseen items. Blue circles denote nonitem entities. Dash circles denote user and item neighborhoods. In this example, a KIG is constructed at first, and then higher hop neighborhood information is aggregated into local neighbors. Finally, the user and item neighborhoods are collected to compute neighborhood interactions. We propose a bi-a… view at source ↗

**Figure 2.** Evaluation stabilizes after sufficient evaluations. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Top-N recommendation results for C-Book. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Top-N recommendation results for Movie-1M. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

**Figure 5.** Figure 5: Top-N recommendation results for A-Book. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

**Figure 6.** Figure 6: Top-N recommendation results for Movie-20M. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

**Figure 7.** Figure 7: Entropy histogram. Note: The x-axis represents the entropy of attention distribution. (a) Ai,j of AAM u365 u735 u1744 u2067 u2685 u4200 u5708 u7521 u7810 u9054 m531 m552 m711 m1031 m2140 m2622 m2908 m2968 m6942 m7155 m62293 m65135 0.00 0.03 0.06 0.09 0.12 0.15 (b) Zi,j…

**Figure 8.** Figure 8: Case study of test case (u46, m3993). Note: In (a)-(d), the y-axis represents neighbors of user 46, and the x-axis represents neighbors of item 3993. AAM: attention aggregation model (Eq. (10)). through user-side and item-side attention networks separately, i.e., Ai,j…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 21 canonical work pages

[1]

Kurt Bollacker, Colin Evans, Praveen Paritosh, Tim Sturge, and Jamie Taylor
[2]

Jie Chen, Tengfei Ma, and Cao Xiao. 2018. FastGCN: fast learning with graph convolutional networks via importance sampling. arXiv preprint arXiv:1801.10247 (2018)

arXiv 2018
[3]

Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, et al
[4]

Hanjun Dai, Yichen Wang, Rakshit Trivedi, and Le Song. 2016. Deep coevolu- tionary network: Embedding user and item features for recommendation. arXiv preprint arXiv:1609.03675 (2016)

arXiv 2016
[5]

Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable feature learning for networks. In SIGKDD. ACM

work page 2016
[6]

Huifeng Guo, Ruiming Tang, Yunming Ye, Zhenguo Li, and Xiuqiang He. 2017. Deepfm: a factorization-machine based neural network for ctr prediction. arXiv preprint arXiv:1703.04247 (2017)

arXiv 2017
[7]

Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. In Advances in Neural Information Processing Systems

2017
[8]

Ruining He and Julian McAuley. 2016. Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative filtering. In WWW. International World Wide Web Conferences Steering Committee

work page 2016

Show all 37 references

[9]

Xiangnan He, Zhenkui He, Jingkuan Song, Zhenguang Liu, Yu-Gang Jiang, and Tat-Seng Chua. 2018. NAIS: Neural Attentive Item Similarity Model for Recom- mendation. TKDE (2018)

2018
[10]

Binbin Hu, Chuan Shi, Wayne Xin Zhao, and Philip S Yu. 2018. Leveraging meta-path based context for top-n recommendation with a neural co-attention model. In SIGKDD. ACM

2018
[11]

Jin Huang, Wayne Xin Zhao, Hong-Jian Dou, Ji-Rong Wen, and Edward Y. Chang
[12]

Xiao Huang, Jundong Li, and Xia Hu. 2017. Label informed attributed network embedding. In WSDM. ACM

2017
[13]

Yuchin Juan, Yong Zhuang, Wei-Sheng Chin, and Chih-Jen Lin. 2016. Field-aware factorization machines for CTR prediction. In RecSys. ACM

2016
[14]

Santosh Kabbur, Xia Ning, and George Karypis. 2013. Fism: factored item simi- larity models for top-n recommender systems. In SIGKDD. ACM

2013
[15]

Thomas N Kipf and Max Welling. 2016. Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907 (2016)

2016 arXiv
[16]

Yehuda Koren. 2008. Factorization meets the neighborhood: a multifaceted collaborative filtering model. In SIGKDD. ACM

2008
[17]

Yankai Lin, Zhiyuan Liu, Maosong Sun, Yang Liu, and Xuan Zhu. 2015. Learning entity and relation embeddings for knowledge graph completion.. In AAAI

2015
[18]

Bryan Perozzi, Rami Al-Rfou, and Steven Skiena. 2014. Deepwalk: Online learning of social representations. In SIGKDD. ACM

2014
[19]

Yanru Qu, Han Cai, Kan Ren, Weinan Zhang, Yong Yu, Ying Wen, and Jun Wang
[20]

Yanru Qu, Bohui Fang, Weinan Zhang, and et al. 2018. Product-based Neural Networks for User Response Prediction over Multi-field Categorical Data. arXiv preprint arXiv:1807.00311 (2018)

2018 arXiv
[21]

Steffen Rendle. 2010. Factorization machines. In ICDM. IEEE

2010
[22]

Jian Tang, Meng Qu, Mingzhe Wang, Ming Zhang, Jun Yan, and Qiaozhu Mei
[23]

Product-based neural networks for user response prediction. In ICDM
[24]

Petar Velickovic, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. 2017. Graph attention networks. arXiv preprint arXiv:1710.10903 (2017)

2017 arXiv
[25]

Hongwei Wang, Jia Wang, Jialin Wang, Miao Zhao, Weinan Zhang, Fuzheng Zhang, Xing Xie, and Minyi Guo. 2017. Graphgan: Graph representation learning with generative adversarial nets. arXiv preprint arXiv:1711.08267 (2017)

2017 arXiv
[26]

Hongwei Wang, Fuzheng Zhang, Jialin Wang, Miao Zhao, Wenjie Li, Xing Xie, and Minyi Guo. 2018. RippleNet: Propagating User Preferences on the Knowledge Graph for Recommender Systems. In CIKM. ACM

2018
[27]

Hongwei Wang, Fuzheng Zhang, Xing Xie, and Minyi Guo. 2018. DKN: Deep Knowledge-Aware Network for News Recommendation. arXiv preprint arXiv:1801.08284 (2018)

2018 arXiv
[28]

Rianne van den Berg, Thomas N Kipf, and Max Welling. 2017. Graph Convolu- tional Matrix Completion. stat 1050 (2017)

2017
[29]

Xiao Yu, Xiang Ren, Yizhou Sun, Quanquan Gu, Bradley Sturt, Urvashi Khandel- wal, Brandon Norick, and Jiawei Han. 2014. Personalized entity recommendation: A heterogeneous information network approach. In WSDM. ACM

2014
[30]

Fuzheng Zhang, Nicholas Jing Yuan, Defu Lian, Xing Xie, and Wei-Ying Ma. 2016. Collaborative knowledge base embedding for recommender systems. InSIGKDD

2016
[31]

Huan Zhao, Quanming Yao, Jianda Li, Yangqiu Song, and Dik Lun Lee. 2017. Meta- graph based recommendation fusion over heterogeneous information networks. In SIGKDD. ACM

2017
[32]

Wayne Xin Zhao, Gaole He, Hongjian Dou, Jin Huang, Siqi Ouyang, and Ji-Rong Wen. 2018. KB4Rec: A Dataset for Linking Knowledge Bases with Recommender Systems. arXiv preprint arXiv:1807.11141 (2018)

2018 arXiv
[33]

Rex Ying, Ruining He, Kaifeng Chen, Pong Eksombatchai, William L Hamilton, and Jure Leskovec. 2018. Graph Convolutional Neural Networks for Web-Scale Recommender Systems. arXiv preprint arXiv:1806.01973 (2018)

2018 arXiv
[2008]

In SIGMOD

Freebase: a collaboratively created graph database for structuring human knowledge. In SIGMOD. AcM
[2015]

Line: Large-scale information network embedding. In WWW. International World Wide Web Conferences Steering Committee
[2016]

In Proceedings of the 1st Workshop on Deep Learning for Recommender Systems

Wide & deep learning for recommender systems. In Proceedings of the 1st Workshop on Deep Learning for Recommender Systems . ACM
[2018]

In SIGIR

Improving Sequential Recommendation with Knowledge-Enhanced Mem- ory Networks. In SIGIR

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Kurt Bollacker, Colin Evans, Praveen Paritosh, Tim Sturge, and Jamie Taylor

[2] [2]

Jie Chen, Tengfei Ma, and Cao Xiao. 2018. FastGCN: fast learning with graph convolutional networks via importance sampling. arXiv preprint arXiv:1801.10247 (2018)

arXiv 2018

[3] [3]

Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, et al

[4] [4]

Hanjun Dai, Yichen Wang, Rakshit Trivedi, and Le Song. 2016. Deep coevolu- tionary network: Embedding user and item features for recommendation. arXiv preprint arXiv:1609.03675 (2016)

arXiv 2016

[5] [5]

Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable feature learning for networks. In SIGKDD. ACM

work page 2016

[6] [6]

Huifeng Guo, Ruiming Tang, Yunming Ye, Zhenguo Li, and Xiuqiang He. 2017. Deepfm: a factorization-machine based neural network for ctr prediction. arXiv preprint arXiv:1703.04247 (2017)

arXiv 2017

[7] [7]

Will Hamilton, Zhitao Ying, and Jure Leskovec. 2017. Inductive representation learning on large graphs. In Advances in Neural Information Processing Systems

2017

[8] [8]

Ruining He and Julian McAuley. 2016. Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative filtering. In WWW. International World Wide Web Conferences Steering Committee

work page 2016