Pith. sign in

REVIEW 3 major objections 5 minor 40 references

Exploiting Text Semantics for Few and Zero Shot Node Classification on Text-attributed Graph

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Adding two text-semantic augmentations to contrastive pre-training beats all 13 baselines on few- and zero-shot node classification for text-attributed graphs.

desk verdict Solid empirical extension of G2P2 with two simple text-side augmentations; the gains look real, but the theoretical appendix should be cut and the retrieval mechanism needs a precision check. read the letter →

arxiv 2505.08168 v1 pith:U4NO55PE submitted 2025-05-13 cs.CL cs.AI

Yuxiang Wang , Xiao Yan , Shiyu Jin , Quanqing Xu , Chuang Hu , Yuanyuan Zhu , Bo Du , Jia Wu
show 1 more author
Jiawei Jiang
This is my paper · ORCID
classification cs.CLcs.AI
keywords text-attributedgraphsfew-shotnodeclassificationzero-shottextsemanticsaugmentationcontrastivelearningprompttuninggraphneuralnetworks
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 is trying to establish that node classification on text-attributed graphs, where each node carries a text description and only a few or no labels are available, can be substantially improved by mining extra text semantics instead of relying only on graph structure. To that end it introduces Text Semantics Augmentation (TSA), which adds two training signals to the standard contrastive alignment of node and text embeddings: matching each node to texts whose embeddings are similar to its own text, and contrasting the node and text with a semantically negated version of the text. On five datasets against thirteen baselines, the paper reports consistent gains for both few-shot and zero-shot classification, with average relative accuracy improvements of 4.6% and 8.8%, respectively. The practical point is that text semantics, not just topology, is a largely untapped supervision source in this setting.

What carries the argument

The load-bearing object is the augmented node-text pair. Positive semantics matching loss $L_{PSM}$ pulls each node embedding toward the $K$ text embeddings retrieved as most similar from a 32K-entry text bank, replacing the single ground-truth text with several reference texts assumed to share the node's label. Negative semantics contrast loss $L_{NSC}$ couples a margin loss $L_{ML}$, which keeps the similarity between a node and its positive text at least a margin above its similarity to a negative text, with a semantics-opposite loss $L_{SO}$ that maximizes the mean squared error between an original text embedding and the embedding of its negated version, where the negation is generated by a learnable negative prompt of $M$ vectors prepended to the raw text. A probability-average inference step for zero-shot classification averages the positive class probability and one minus the negative-class probability, so the negative text encoder contributes an independent predictive signal. These mechanisms add text-semantic supervision to the base InfoNCE contrastive loss $L_{CL}$.

What would settle it

On any of the five datasets, compute the label agreement between each node and its K most similar text-bank texts; if agreement at K=1 is near the base class prior, the positive semantics matching term cannot be doing semantic work and the reported gain would need another explanation.

Watch

Extended reading notes

Core claim

TSA claims that the weak spot of prior end-to-end methods is that they contrast only the single given node-text pair, leaving most of the semantic information in the text unused. The paper's central discovery is that two simple augmentations close this gap: positive semantics matching retrieves the K most similar text embeddings from a rolling text bank and pulls the node embedding toward them, while negative semantics contrast prepends a learnable negative prompt to each text to produce an opposite-semantics text and then uses a margin loss plus a semantics-opposite loss to push the node and original text away from it. With these terms added to the base contrastive loss, TSA outperforms all 13 baselines on Cora, Fitness, M.I., Industrial, and Art; the average relative gains over the best baseline are 4.6% accuracy and 6.9% F1 in few-shot, and 8.8% accuracy and 9.3% F1 in zero-shot. The gain is larger in zero-shot, which the authors attribute to the negative contrast supplying the semantics that labeled data would otherwise provide.

Load-bearing premise

The claim's main load-bearing premise is that texts retrieved from the text bank by embedding similarity actually share the target node's label; the paper never measures that retrieval precision, sets K to 1 in its main experiments, and its proof that a learnable negative prompt captures true negation is approximate.

Editorial extensions

If this is right

  • If the reported gains hold, state-of-the-art few- and zero-shot node classification on text-attributed graphs should come from augmenting contrastive pre-training with similar-text matching rather than from scaling graph augmentations or model size.
  • Because gains are larger in zero-shot than few-shot, negative semantics contrast is most valuable precisely when no labels are available, which makes the technique a natural fit for cold-start settings.
  • The ablations show that adding negative semantics contrast helps zero-shot but hurts few-shot classification, so deployment should activate the negative term only for zero-shot.
  • TSA reaches the accuracy of the best previous method with fewer labeled shots and similar pre-training and inference time, so the accuracy gains do not come at a major efficiency cost.

Reading between the lines

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

  • A direct test of retrieval precision in the text bank would separate two explanations of the gain: semantic matching versus merely having extra positive pairs; the paper does not report this precision.
  • If retrieval remains accurate at higher K, the text bank could double as a pseudo-label source for semi-supervised or self-training pipelines on text-attributed graphs.
  • The learnable negative prompt may transfer to other contrastive text models as a cheap way to add negation awareness, for example in question answering or stance detection, though the paper only evaluates node classification.
  • The approximate proof of Theorem 1 leaves open whether a hand-crafted negation corpus would give the same zero-shot gains; comparing the learnable prompt against a manual negation bank is a direct extension.
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.

Referee Report

3 major / 5 minor

Summary. The paper proposes Text Semantics Augmentation (TSA) for few-shot and zero-shot node classification on text-attributed graphs. TSA extends the G2P2 contrastive training pipeline with two text-level augmentations: positive semantics matching, which retrieves similar texts from a FIFO text bank and treats them as additional positive node-text pairs, and negative semantics contrast, which uses learnable negative prompts and an independently trained negative text encoder to push node/text embeddings away from semantically negated descriptions. The authors also introduce a probability-average inference strategy for the zero-shot setting. They report experiments on five datasets against thirteen baselines, with consistent accuracy and F1 gains over the best baseline, averaging 4.6% and 6.9% relative improvement for few-shot and 8.8% and 9.3% for zero-shot classification. A theoretical claim (Theorem 1) in Appendix A asserts that learnable negative prompts have higher information entropy than a lower bound of hand-crafted negative prompts.

Significance. If the empirical results hold, TSA is a simple and general augmentation scheme that improves a state-of-the-art co-trained GNN/LM model on standard TAG benchmarks, with especially large gains in zero-shot classification. The experimental protocol is solid on its face: five datasets, multiple baseline families, five repeated runs with means and standard deviations, and ablations that decompose the contribution of each loss term. The efficiency comparison with G2P2 is also useful. However, the paper's only theoretical justification, Theorem 1, is not rigorously established, and the mechanism behind the positive semantics matching gain is not directly validated. These issues do not overturn the empirical claim, but they do mean the paper currently overstates what it has shown.

major comments (3)
  1. [Appendix A, Theorem 1 (Eqs. 11-15)] The proof of Theorem 1 is not valid. Equations (11) and (12) treat sums of per-token probabilities as entropies without defining a proper probability space, and the factorization into 'negation words' and 'raw text' assumes an independence that is not justified. Equation (13) is stated as an approximation ('approximately equal to') rather than a lower bound, so the subsequent comparison to a lower bound is not meaningful. The step dropping the raw-text entropy term, -sum_j P(x_j) log P(x_j) ≈ 0, is unjustified, and the Maximum Entropy Theorem is applied to a non-normalized sum. Consequently, H(h) ≥ LowerBound(H(Xneg)) is not established. Since this theorem is presented as formal support for learnable negative prompts, the authors should either provide a rigorous proof or explicitly recharacterize this argument as a heuristic motivation.
  2. [Section 3.2, Eq. (3)] The positive semantics matching loss assumes that texts retrieved from the text bank by embedding similarity share the target node's label. This label-sharing prior is never measured. The paper does not report retrieval precision, same-label rate, or any qualitative analysis of the retrieved texts, and with the FIFO bank capacity of 32K covering only recent batches and K=1 in the main experiments, the retrieved text may frequently be irrelevant or noisy. Without this measurement, L_PSM could be acting as a self-supervised noise term rather than a semantic augmentation, so the mechanism behind the reported gains is unverified. I ask the authors to add a retrieval-precision analysis and an ablation that replaces retrieved texts with random texts; this would directly test whether the positive matching mechanism is responsible for the improvements.
  3. [Section 4.1 vs. Section 4.3 / Figure 5] There is an internal inconsistency about the number of similar texts K. Section 4.1 states that K is set to 1 for the main experiments, while the hyperparameter analysis in Section 4.3 and Figure 5 concludes that accuracy first increases and then decreases with K, with the optimum at an intermediate value. If intermediate K values are better, the main tables may under-report the method's performance, and the choice of K=1 is unexplained. Please clarify the actual K values in Figure 5, report results with the tuned K in the main tables, or justify why K=1 is used despite the stated trend.
minor comments (5)
  1. [Section 3.3] The phrase 'probability-average stragety' contains a typo; it should be 'strategy'.
  2. [Figure 5 caption] The caption reads 'M.I. anf Industrial'; 'anf' should be 'and'.
  3. [Section 4.3 / Table 3] The text says 'all augmentations are effective in improving accuracy,' but for few-shot classification adding NSC to LCL+PSM slightly decreases accuracy (e.g., M.I. 87.91 to 87.80). The wording should be adjusted to state that NSC is beneficial only in the zero-shot setting.
  4. [Equations (2) and (3)] In both contrastive losses, the denominator sums over j≠i and excludes the positive term(s). This is not the standard InfoNCE form; if this is intentional, please clarify why the positive is excluded, otherwise the equations should be corrected.
  5. [Section 4.2 / Tables 1-2] The paper reports means and standard deviations but no statistical significance tests. Given that the reported gains are often larger than the standard deviations, this is not a blocking issue, but a paired test across the five runs would strengthen the claim of consistent improvement.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the central accuracy claims rest on external-baseline comparisons, and the paper's self-citations and Theorem 1 are not load-bearing in a circular sense.

full rationale

The derivation chain is not circular. TSA's two augmentations, L_PSM (Eq. 3) and L_NSC (Eqs. 6-7), are new training objectives rather than restatements of the evaluation metric or of the label assignment; the reported few-shot and zero-shot numbers in Tables 1 and 2 are obtained by comparing the resulting model against 13 external baselines, so the central claim is empirically self-contained. The label-sharing prior behind positive semantics matching is an assumption about retrieval quality, not a definitional equivalence: the paper does not measure retrieval precision, and K=1 is used in the main experiments, but this is a mechanism-validation gap, not a circular reduction. Theorem 1 (Section 3.2 and Appendix A) is the weakest point: the proof approximately drops the raw-text entropy, assumes high-dimensional prompt vectors are necessarily more dispersed, and then invokes maximum entropy, so the inequality H(h) >= LowerBound(H(Xneg)) is not rigorously established. However, that theorem is a post-hoc justification for the learnable negative prompt and does not feed back into the accuracy numbers; it is a rigor concern rather than a circular one. The paper's self-citations (Wang et al. 2024a,b; Wang et al. 2025; Liang et al. 2025; Zhang et al. 2025) appear only in related-work context and do not carry the central claim. No fitted parameter is renamed as a prediction, and no uniqueness theorem from the authors is invoked to forbid alternatives. Hence the score is 1, reflecting only the minor unsupported Theorem 1 justification.

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

The method introduces no new physical entities; the negative text encoder is a model component, and the learnable negative prompt is a trainable vector sequence. The main additions are five tuned hyperparameters and two domain assumptions.

free parameters (5)
  • alpha (NSC loss weight) = 0 for few-shot, 0.5 for zero-shot
    Chosen per task in Section 3.2 and justified by the ablation in Table 3.
  • K (number of similar texts) = 1
    Set to 1 in Section 4.1; Figure 5 shows accuracy peaks at an intermediate K, implying it is a tuned hyperparameter.
  • Text bank capacity = 32K
    Set to 32K embeddings to balance memory and retrieval quality (Section 3.2, 4.1).
  • Negative prompt length M = 16
    Set to 16 in Section 4.1; no sensitivity analysis is provided.
  • Margin m = 1
    Set to 1 in Eq. (6) and Section 4.1; no sensitivity analysis is provided.
assumptions (4)
  • domain assumption Texts with similar embeddings in the text bank share the target node's label
    Central prior for the positive semantics matching loss in Eq. (3); the paper provides no retrieval-precision analysis.
  • ad hoc to paper Learnable negative prompt vectors can capture semantic negation of the raw text
    The paper's Theorem 1 claims this, but the appendix proof is approximate and ignores raw-text entropy terms.
  • standard math InfoNCE random negatives approximately represent the full embedding distribution
    Standard contrastive learning assumption inherited from He et al. 2020 and used in Eq. (2).
  • domain assumption TAG nodes have a strict one-to-one correspondence with their text descriptions
    Stated in Section 2 preliminaries and used to define node-text pairs for contrastive learning.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Exploiting Text Semantics for Few and Zero Shot Node Classification on Text-attributed Graph." pith.science (2026). https://pith.science/paper/U4NO55PE

@misc{pith2026250508168,
  author       = {Pith},
  title        = {Pith review of: Exploiting Text Semantics for Few and Zero Shot Node Classification on Text-attributed Graph},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U4NO55PE}},
  note         = {Machine review of arXiv:2505.08168}
}
read the original abstract

Text-attributed graph (TAG) provides a text description for each graph node, and few- and zero-shot node classification on TAGs have many applications in fields such as academia and social networks. Existing work utilizes various graph-based augmentation techniques to train the node and text embeddings, while text-based augmentations are largely unexplored. In this paper, we propose Text Semantics Augmentation (TSA) to improve accuracy by introducing more text semantic supervision signals. Specifically, we design two augmentation techniques, i.e., positive semantics matching and negative semantics contrast, to provide more reference texts for each graph node or text description. Positive semantic matching retrieves texts with similar embeddings to match with a graph node. Negative semantic contrast adds a negative prompt to construct a text description with the opposite semantics, which is contrasted with the original node and text. We evaluate TSA on 5 datasets and compare with 13 state-of-the-art baselines. The results show that TSA consistently outperforms all baselines, and its accuracy improvements over the best-performing baseline are usually over 5%.

Figures

Figures reproduced from arXiv: 2505.08168 by the authors.

Figure 1
Figure 1. The contrastive loss of G2P2 (a) and two semantic aug [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overview of TSA. • Negative text encoder ψ neg. This maintains the same ar￾chitecture as the text encoder, with the difference that we train it independently with a negative prompt to generate the negative text presentation t neg . To effectively train the above encoders, we design two novel loss functions: positive semantics matching loss and negative semantics contrast loss, which can assist the pre￾training m… view at source ↗
Figure 3
Figure 3. The illustration of probability-average. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The time cost comparison of pre-training and prompting [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The comparison of the number of similar texts and the [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: The accuracy comparison for our TSA and G2P2 in the [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 29 canonical work pages

  1. [1]

    Label-free node classification on graphs with large language models (llms)

    [Chen et al., 2023] Zhikai Chen, Haitao Mao, Hongzhi Wen, Haoyu Han, Wei Jin, Haiyang Zhang, Hui Liu, and Jiliang Tang. Label-free node classification on graphs with large language models (llms). arXiv preprint arXiv:2310.04668,

    arXiv 2023

  2. [7]

    Graph neural networks for recommender system

    [Gao et al., 2022] Chen Gao, Xiang Wang, Xiangnan He, and Yong Li. Graph neural networks for recommender system. In Proceedings of the 15th ACM International Conference on Web Search and Data Mining, pages 1623– 1625,

    work page 2022

  3. [10]

    Momentum contrast for un- supervised visual representation learning

    [He et al., 2020] Kaiming He, Haoqi Fan, Yuxin Wu, Sain- ing Xie, and Ross Girshick. Momentum contrast for un- supervised visual representation learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 9729–9738,

    work page 2020

  4. [11]

    Gpt-gnn: Genera- tive pre-training of graph neural networks

    [Hu et al., 2020] Ziniu Hu, Yuxiao Dong, Kuansan Wang, Kai-Wei Chang, and Yizhou Sun. Gpt-gnn: Genera- tive pre-training of graph neural networks. In Proceed- ings of the 26th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages 1857– 1867,

    work page 2020

  5. [12]

    Prompt-based node feature extractor for few-shot learning on text-attributed graphs

    [Huang et al., 2023] Xuanwen Huang, Kaiqiao Han, Dezheng Bao, Quanjin Tao, Zhisheng Zhang, Yang Yang, and Qi Zhu. Prompt-based node feature extractor for few-shot learning on text-attributed graphs. arXiv preprint arXiv:2309.02848,

    arXiv 2023

  6. [13]

    Semi-supervised classification with graph convolutional networks

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

    arXiv 2016

  7. [16]

    Roberta: a robustly optimized bert pretraining approach

    [Liu et al., 2019] Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. Roberta: a robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692,

    arXiv 2019

  8. [18]

    Few-shot node classification on attributed networks with graph meta-learning

    [Liu et al., 2022] Yonghao Liu, Mengyu Li, Ximing Li, Fausto Giunchiglia, Xiaoyue Feng, and Renchu Guan. Few-shot node classification on attributed networks with graph meta-learning. In Proceedings of the 45th interna- tional ACM SIGIR Conference on Research and Develop- ment in Information Retrieval, pages 471–481,

    work page 2022

Show all 40 references
  1. [19]

    Graphprompt: Unifying pre-training and downstream tasks for graph neural networks

    [Liu et al., 2023] Zemin Liu, Xingtong Yu, Yuan Fang, and Xinming Zhang. Graphprompt: Unifying pre-training and downstream tasks for graph neural networks. In Proceed- ings of the ACM Web Conference 2023 , pages 417–428,

    2023

  2. [20]

    Automat- ing the construction of internet portals with machine learn- ing

    [McCallum et al., 2000] Andrew Kachites McCallum, Ka- mal Nigam, Jason Rennie, and Kristie Seymore. Automat- ing the construction of internet portals with machine learn- ing. Information Retrieval, 3:127–163,

    2000

  3. [23]

    Graphgpt: Graph instruction tuning for large language models

    [Tang et al., 2024] Jiabin Tang, Yuhao Yang, Wei Wei, Lei Shi, Lixin Su, Suqi Cheng, Dawei Yin, and Chao Huang. Graphgpt: Graph instruction tuning for large language models. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Infor- mat...

    2024

  4. [24]

    Attention is all you need

    [Vaswani et al., 2017] Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in Neural Information Processing Sys- tems, 30,

    2017

  5. [25]

    Graph attention networks

    [Veliˇckovi´c et al., 2017] Petar Veliˇckovi´c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. Graph attention networks. arXiv preprint arXiv:1710.10903,

    2017 arXiv

  6. [26]

    Deep graph infomax

    [Veliˇckovi´c et al., 2018] Petar Veli ˇckovi´c, William Fedus, William L Hamilton, Pietro Li `o, Yoshua Bengio, and R Devon Hjelm. Deep graph infomax. arXiv preprint arXiv:1809.10341,

    2018 arXiv

  7. [27]

    Generative and contrastive paradigms are complementary for graph self-supervised learning

    [Wang et al., 2024a] Yuxiang Wang, Xiao Yan, Chuang Hu, Quanqing Xu, Chuanhui Yang, Fangcheng Fu, Wentao Zhang, Hao Wang, Bo Du, and Jiawei Jiang. Generative and contrastive paradigms are complementary for graph self-supervised learning. In 2024 IEEE 40th International Confere...

    2024

  8. [28]

    TESA: A trajectory and semantic-aware dy- namic heterogeneous graph neural network

    [Wang et al., 2025] Xin Wang, Jiawei Jiang, Xiao Yan, and Qiang Huang. TESA: A trajectory and semantic-aware dy- namic heterogeneous graph neural network. In Proceed- ings of the ACM on Web Conference , pages 1305–1315. ACM,

    2025

  9. [29]

    Aug- menting low-resource text classification with graph- grounded pre-training and prompting

    [Wen and Fang, 2023] Zhihao Wen and Yuan Fang. Aug- menting low-resource text classification with graph- grounded pre-training and prompting. In Proceedings of the 46th International ACM SIGIR Conference on Re- search and Development in Information Retrieval , pages 506–516,

    2023

  10. [30]

    Infogcl: Information-aware graph contrastive learning.Advances in Neural Information Processing Systems, 34:30414–30425,

    [Xu et al., 2021] Dongkuan Xu, Wei Cheng, Dongsheng Luo, Haifeng Chen, and Xiang Zhang. Infogcl: Information-aware graph contrastive learning.Advances in Neural Information Processing Systems, 34:30414–30425,

    2021

  11. [31]

    A comprehensive study on text-attributed graphs: Benchmarking and re- thinking

    [Yan et al., 2023] Hao Yan, Chaozhuo Li, Ruosong Long, Chao Yan, Jianan Zhao, Wenwen Zhuang, Jun Yin, Peiyan Zhang, Weihao Han, Hao Sun, et al. A comprehensive study on text-attributed graphs: Benchmarking and re- thinking. Advances in Neural Information Processing Sys- tems, ...

    2023

  12. [32]

    Oceanbase: a 707 million tpmc dis- tributed relational database system

    [Yang et al., 2022] Zhenkun Yang, Chuanhui Yang, Fusheng Han, Mingqiang Zhuang, Bing Yang, Zhifeng Yang, Xi- aojun Cheng, Yuzhong Zhao, Wenhui Shi, Huafeng Xi, Huang Yu, Bin Liu, Yi Pan, Boxue Yin, Junquan Chen, and Quanqing Xu. Oceanbase: a 707 million tpmc dis- tributed rela...

    2022

  13. [33]

    Oceanbase paetica: A hybrid shared-nothing/shared- everything database for supporting single machine and dis- tributed cluster

    [Yang et al., 2023] Zhifeng Yang, Quanqing Xu, Shanyan Gao, Chuanhui Yang, Guoping Wang, Yuzhong Zhao, Fanyu Kong, Hao Liu, Wanhong Wang, and Jinliang Xiao. Oceanbase paetica: A hybrid shared-nothing/shared- everything database for supporting single machine and dis- tributed c...

    2023

  14. [34]

    Graph convolutional networks for text classification

    [Yao et al., 2019] Liang Yao, Chengsheng Mao, and Yuan Luo. Graph convolutional networks for text classification. In Proceedings of the AAAI Conference on Artificial Intel- ligence, volume 33, pages 7370–7377,

    2019

  15. [35]

    Graph con- trastive learning with augmentations

    [You et al., 2020] Yuning You, Tianlong Chen, Yongduo Sui, Ting Chen, Zhangyang Wang, and Yang Shen. Graph con- trastive learning with augmentations. Advances in Neural Information Processing Systems, 33:5812–5823,

    2020

  16. [36]

    Enhancing social recom- mendation with adversarial graph convolutional networks

    [Yu et al., 2020] Junliang Yu, Hongzhi Yin, Jundong Li, Min Gao, Zi Huang, and Lizhen Cui. Enhancing social recom- mendation with adversarial graph convolutional networks. IEEE Transactions on Knowledge and Data Engineering , 34(8):3727–3739,

    2020

  17. [37]

    Graphtranslator: Align- ing graph model to large language model for open-ended tasks

    [Zhang et al., 2024] Mengmei Zhang, Mingwei Sun, Peng Wang, Shen Fan, Yanhu Mo, Xiaoxiao Xu, Hong Liu, Cheng Yang, and Chuan Shi. Graphtranslator: Align- ing graph model to large language model for open-ended tasks. In Proceedings of the ACM on Web Conference 2024, pages 1003–1014,

    2024

  18. [38]

    Tgraph: A tensor-centric graph processing framework

    [Zhang et al., 2025] Yongliang Zhang, Yuanyuan Zhu, Hao Zhang, Congli Gao, Yuyang Wang, Guojing Li, Tianyang Xu, Ming Zhong, Jiawei Jiang, Tieyun Qian, Chenyi Zhang, and Jeffrey Xu Yu. Tgraph: A tensor-centric graph processing framework. Proc. ACM Manag. Data , 3(1):81:1–81:27,

    2025

  19. [39]

    Pre-training and prompting for few-shot node classification on text- attributed graphs

    [Zhao et al., 2024] Huanjing Zhao, Beining Yang, Yukuo Cen, Junyu Ren, Chenhui Zhang, Yuxiao Dong, Evgeny Kharlamov, Shu Zhao, and Jie Tang. Pre-training and prompting for few-shot node classification on text- attributed graphs. arXiv preprint arXiv:2407.15431,

    2024 arXiv

  20. [40]

    Conditional prompt learning for vision-language models

    [Zhou et al., 2022] Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. Conditional prompt learning for vision-language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16816–16825, 2022

    2022

  21. [2000]

    Graph-based anomaly detection

    [Noble and Cook, 2003] Caleb C Noble and Diane J Cook. Graph-based anomaly detection. In Proceedings of the ninth ACM SIGKDD International Conference on Knowl- edge Discovery and Data Mining, pages 631–636,

    2003

  22. [2003]

    Gppt: Graph pre-training and prompt tuning to generalize graph neural networks

    [Sun et al., 2022] Mingchen Sun, Kaixiong Zhou, Xin He, Ying Wang, and Xin Wang. Gppt: Graph pre-training and prompt tuning to generalize graph neural networks. InPro- ceedings of the 28th ACM SIGKDD Conference on Knowl- edge Discovery and Data Mining, pages 1717–1727,

    2022

  23. [2016]

    Towards scalable and deep graph neural networks via noise masking

    [Liang et al., 2025] Yuxuan Liang, Wentao Zhang, Zeang Sheng, Ling Yang, Quanqing Xu, Jiawei Jiang, Yunhai Tong, and Bin Cui. Towards scalable and deep graph neural networks via noise masking. In Sponsored by the Association for the Advancement of Artificial Intelligence, page...

    2025

  24. [2017]

    Palf: Replicated write-ahead logging for distributed databases

    [Han et al., 2024] Fusheng Han, Hao Liu, Bin Chen, De- bin Jia, Jianfeng Zhou, Xuwang Teng, Chuanhui Yang, Huafeng Xi, Wei Tian, Shuning Tao, Sen Wang, Quan- qing Xu, and Zhenkun Yang. Palf: Replicated write-ahead logging for distributed databases. Proc. VLDB Endow. , 17(12):3...

    2024

  25. [2018]

    Universal prompt tuning for graph neural networks

    [Fang et al., 2024] Taoran Fang, Yunchao Zhang, Yang Yang, Chunping Wang, and Lei Chen. Universal prompt tuning for graph neural networks. Advances in Neural In- formation Processing Systems, 36,

    2024

  26. [2019]

    Relative and absolute location embed- ding for few-shot node classification on graph

    [Liu et al., 2021] Zemin Liu, Yuan Fang, Chenghao Liu, and Steven CH Hoi. Relative and absolute location embed- ding for few-shot node classification on graph. In Pro- ceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 4267–4275,

    2021

  27. [2020]

    Graph neural network-based anomaly detection in multivariate time series

    [Deng and Hooi, 2021] Ailin Deng and Bryan Hooi. Graph neural network-based anomaly detection in multivariate time series. In Proceedings of the AAAI Conference on Ar- tificial Intelligence, volume 35, pages 4027–4035,

    2021

  28. [2021]

    Bert: pre-training of deep bidirectional transformers for language understand- ing

    [Devlin et al., 2018] Jacob Devlin, Ming-Wei Chang, Ken- ton Lee, and Kristina Toutanova. Bert: pre-training of deep bidirectional transformers for language understand- ing. arXiv preprint arXiv:1810.04805,

    2018 arXiv

  29. [2022]

    Inductive representation learning on large graphs

    [Hamilton et al., 2017] Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. Advances in Neural Information Processing Sys- tems, 30,

    2017

  30. [2023]

    Exploring the potential of large language models (llms) in learning on graphs

    [Chen et al., 2024] Zhikai Chen, Haitao Mao, Hang Li, Wei Jin, Hongzhi Wen, Xiaochi Wei, Shuaiqiang Wang, Dawei Yin, Wenqi Fan, Hui Liu, et al. Exploring the potential of large language models (llms) in learning on graphs. ACM SIGKDD Explorations Newsletter, 25(2):42–61,

    2024

  31. [2024]

    Debiased contrastive learning

    [Chuang et al., 2020] Ching-Yao Chuang, Joshua Robinson, Yen-Chen Lin, Antonio Torralba, and Stefanie Jegelka. Debiased contrastive learning. Advances in Neural Infor- mation Processing Systems, 33:8765–8775,

    2020

  32. [2025]

    Prototypical graph contrastive learning

    [Lin et al., 2022] Shuai Lin, Chen Liu, Pan Zhou, Zi-Yuan Hu, Shuojia Wang, Ruihui Zhao, Yefeng Zheng, Liang Lin, Eric Xing, and Xiaodan Liang. Prototypical graph contrastive learning. IEEE Transactions on Neural Net- works and Learning Systems, 35(2):2747–2758,

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.