SGPT: Few-Shot Prompt Tuning for Signed Graphs

Sima Qing; Wenjie Zhang; Xiaoyang Wang; Zian Zhai

REVIEW 4 major objections 6 minor 53 references

SGPT: Few-Shot Prompt Tuning for Signed Graphs

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read The paper proposes SGPT, a prompt-tuning framework that transfers pre-trained unsigned GNNs to few-shot signed graph learning.

desk verdict First prompt-tuning framework for signed graphs with a sensible design and strong results, but the balance-theory disentanglement is under-validated and the evaluation has protocol issues; worth serious refereeing but needs revision. read the letter →

arxiv 2412.12155 v2 pith:DSGIKHTL submitted 2024-12-11 cs.LG cs.AI

Zian Zhai , Sima Qing , Xiaoyang Wang , Wenjie Zhang This is my paper · ORCID

classification cs.LGcs.AI

keywords signedgraphsprompttuningfew-shotlearninggraphneuralnetworksbalancetheorylinksignpredictionnodeclassificationpre-training

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

Pre-trained graph neural networks are usually built for unsigned graphs, where links only mean connection, while signed graph tasks need to know whether a link is trust or distrust. SGPT claims that a frozen unsigned GNN can be reused for both node classification and link sign prediction with very few labels, as long as the signed graph is first repackaged by templates that make it look like the pre-training data. If true, this would let practitioners skip training specialized signed GNNs and instead adapt abundant unsigned graph models with only a few thousand tunable prompt parameters. The paper reports consistent gains over supervised signed models and prior graph-prompt methods on seven datasets under few-shot settings.

What carries the argument

The load-bearing mechanism is the three-channel graph template built from balance theory. With $A^1_P = A^+$ and $A^1_N = A^-$, each further hop is computed as $A^k_P = I(A^{k-1}_P A^1_P + A^{k-1}_N A^1_N)$ and $A^k_N = I(A^{k-1}_P A^1_N + A^{k-1}_N A^1_P)$, where $I(\cdot)$ binarizes non-zero entries; this encodes whether paths have an even or odd number of negative links. The positive, negative, and topological samples are encoded in parallel by the frozen pre-trained GNN, aligned by channel-specific feature prompts and fused by a semantic prompt, so the whole pipeline behaves like the pre-training link-prediction task.

What would settle it

Compute the structural balance level of the seven datasets, or split any signed graph into high-balance and low-balance subgraphs, and check whether SGPT's AUC drops as contradictory paths become more common; if performance stays flat even when balance is low, the channel decomposition itself is not what drives the reported gains.

Watch

Extended reading notes

Core claim

The central claim is that the structural gap between unsigned pre-training and signed downstream tasks can be closed by a balance-theory graph template that splits a signed graph into three channels: positive, negative, and purely topological. Multi-hop positive and negative adjacencies are built by recursive rules that encode the balance-theoretic sign of a path, so each channel contains links of one consistent semantics and can be fed into an unsigned GNN without violating its homophily assumption. A task template then rewrites both node classification and link sign prediction as prototype-based link prediction, matching the pre-training objective. Lightweight feature prompts adjust each channel's input space, and a bottleneck-adapter semantic prompt fuses the channels task-adaptively. The paper claims this is the first prompt-tuning framework to transfer pre-trained unsigned GNNs to few-shot signed tasks and that it outperforms existing supervised SGNNs and unsigned prompt-tuning baselines on seven benchmark datasets.

Load-bearing premise

The method assumes balance theory holds on the test graphs, so the sign of any multi-hop relationship is determined by whether the number of negative links along the path is even, and contradictory paths can be safely binarized into positive or negative channels.

Editorial extensions

If this is right

Signed graph node classification and link sign prediction can in principle be solved without training a dedicated signed GNN, reusing a frozen unsigned backbone instead.
Because only prompts are tuned, the downstream phase needs very few labels and far fewer tunable parameters than supervised fine-tuning, which helps in label-scarce industrial settings.
The same template and prompt design carries across different pre-trained backbones (GCN, GAT, GIN), so the method is not tied to one encoder architecture.
The task template makes two seemingly different tasks share one objective, so gains on one signed task can transfer to the other.
Balance-theoretic multi-hop relationships give the method a way to incorporate global structure beyond direct signed neighbors.

Reading between the lines

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

Editorial inference: the method's success should depend on how balanced the test graphs actually are; datasets with many contradictory paths between the same pair are where the binarizing indicator function in the recursive adjacency construction is least reliable.
Editorial inference: because the indicator function discards path multiplicity, a testable extension is to weight channels by counts or confidence of balanced versus unbalanced paths rather than binarizing them, which may improve robustness on noisy signed networks.
Editorial inference: the framework suggests a practical recipe beyond the paper's benchmarks: take any pre-trained unsigned GNN, add these templates, and probe it on directed or weighted signed graphs, provided the balance-theoretic sign rule still applies.

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,

The short version: this is the first paper to bring prompt tuning to signed graphs, and the overall idea is sound. But the core contribution -- the balance-theory graph template -- rests on an assumption the paper never checks, and the experimental protocol has a couple of unfair comparisons. I'd send it to a serious referee, but I wouldn't take the main claim at face value yet.

What's genuinely new: SGPT is the first to adapt pre-trained unsigned GNNs to few-shot signed tasks. The graph template decomposes the signed graph into positive, negative, and topological channels using balance-theory path parity. The task template unifies node classification and link sign prediction into link prediction, which is a neat trick. Feature prompts and semantic prompts are lightweight and parameter-efficient. The experiments cover seven datasets and show consistent gains over supervised SGNNs, fine-tuning, and unsigned prompt-tuning baselines. The ablation study shows each component contributes, and the complexity analysis is honest.

Where it gets soft. The stress-test note is on target. Equations 4-5 use an indicator function that binarizes nonzero entries. If a node pair is connected by both balanced and unbalanced paths -- common in real signed networks -- that pair goes into both the positive and negative channels. The claimed 'consistent type of internal link semantics' doesn't hold. The paper cites SGCL for the prevalence of balanced structures but reports no balance statistics for any dataset and runs no synthetic experiments with controlled balance. So the mechanism that supposedly justifies the template is unverified where it matters most. The strong ablation of the graph template (Variant-1) shows it helps, but doesn't distinguish balance-theoretic disentanglement from simply adding 2-hop edges.

There's also a copy-paste error in Table 2: GraphPrompt+ has the same value (75.11±0.24) for Epinions and Wikipedia-RfA. The LSP protocol gives signed models the signs of 30% of links for message-passing, while unsigned baselines see only topology; that tilts the comparison. No significance testing, and the pre-training graph and prompt hyperparameters (basis count, bottleneck dimension) aren't specified for the main runs.

Bottom line: this is a genuine first step with a sensible design, but the central claim is under-supported. The paper deserves a serious referee, but the authors need to address the ambiguity issue, fix the protocol, and add statistical tests before the 'universal superiority' claim is credible.

Referee Report

4 major / 6 minor

Summary. The paper proposes SGPT (Signed Graph Prompt Tuning), a framework that adapts pre-trained unsigned GNNs to few-shot signed graph tasks (node classification and link sign prediction). SGPT introduces a graph template based on balance theory that separates multi-hop relationships into positive, negative, and topological channels; a task template that unifies downstream tasks into a link-prediction form; channel-specific feature prompts; and a semantic-prompt adapter that fuses channel embeddings. The authors evaluate on seven signed-network datasets against supervised SGNNs, pre-train/fine-tune methods, and graph prompt-tuning baselines, reporting ROC-AUC improvements across tasks and shots.

Significance. If the reported results are valid, SGPT would be a practically useful contribution to label-scarce signed graph learning, showing how abundant unsigned graph pre-training can be transferred to signed tasks without retraining the backbone. The paper is well structured, provides ablations that isolate the proposed components, includes sensitivity analyses for the hop number and prompt basis size, and offers complexity analysis. However, the main empirical claim is currently weakened by an evaluation protocol that gives signed-aware models extra sign information, and the core disentanglement property of the graph template is not validated under ambiguous multi-hop paths. These issues are addressable but must be fixed before the contribution can be assessed fairly.

major comments (4)

[§5.1 (Implementations)] The link sign prediction protocol is not fair across methods. The paper states that for the 30% message-passing links, signed graph models (including SGPT, SGCN, and SDGNN) know the link signs when generating embeddings or applying the graph template, while unsigned models treat these links as unsigned. This gives SGPT and other signed baselines access to 30% of the ground-truth signs in addition to the few-shot training labels, while unsigned prompt baselines (GPPT, GraphPrompt, GPF+, etc.) do not receive that information. The observed advantage over unsigned models may therefore reflect this extra supervision rather than the proposed templates or prompts. Please rerun experiments under a controlled protocol in which all methods see the same sign information (or none do), or add an ablation where unsigned models also receive the 30% signs as an explicit two-channel edge feature, so the contribution of the graph template and prompts can be isolated.
[§4.3, Eqs. (4)–(5)] The graph template claims to disentangle mixed node relationships into channels with consistent link semantics, but the construction via the indicator function assumes that each node pair has a unique balance-theoretic sign at each hop. If a pair (i,j) is connected by both a balanced and an unbalanced k-hop path, then both A^k_P(i,j)=1 and A^k_N(i,j)=1 after applying I(·), so the same edge is inserted into both the positive and negative channels. The paper asserts that balanced structures are prevalent in real-world networks but reports no balance statistics (e.g., fraction of balanced triads or fraction of node pairs with mixed-sign paths) for any of the seven datasets, and provides no synthetic experiments with controlled balance levels. Please measure the prevalence of ambiguous path-sign pairs in the datasets or include a controlled synthetic study; if ambiguity is common, explain why the method remains effective or modify the template (e.g., using path counts or sign voting) to preserve disentanglement.
[§5.2, Table 2 and Fig. 3] The paper claims that SGPT "significantly outperforms" existing methods, but no statistical significance testing is reported. Several differences in Table 2 are within one standard deviation (e.g., WikiEditor: SGPT 54.17±5.51 vs. GPF+ 52.61±5.28; WikiElec: GraphPrompt+ 51.23±9.11 vs. SGCN 51.06±1.52). Since the results are aggregated over 100 random tasks, paired significance tests (e.g., Wilcoxon signed-rank or paired t-test) should be reported for all main comparisons, and the wording should be adjusted to the actual statistical evidence.
[§5.1 and §4.2] Pre-training and prompt-tuning details are insufficient for reproducibility. The paper does not specify which unsigned graph(s) were used for pre-training, their size and domain, the node feature transformation, the number of pre-training epochs, the number of prompt-tuning epochs, batch size, or how the class prototypes in Eqs. (10)–(11) are initialized and updated. These details are necessary to reproduce the reported numbers and to allow the community to build on the method. Please add a reproducible experimental setup description or an appendix with these hyperparameters.

minor comments (6)

[Table 2] The GraphPrompt+ entry for Wikipedia-RfA (75.11±0.24) is identical to the Epinions entry (75.11±0.24) and appears to be a copy-paste error; please correct it.
[Table 2 and §5.2] The dataset name "Wikiedia-Editor" in Table 2 should be "Wikipedia-Editor", and "GraphPromp+" in Section 5.2 should be "GraphPrompt+".
[§1] "preserving instinct graph structures" should likely be "preserving intrinsic graph structures".
[References] References [9] and [10] are duplicates (both cite "Signed graph attention networks" by Huang et al.), and references [23] and [24] are also duplicates (both cite the same Mercado et al. paper); please consolidate.
[§1] The claim of being "the first study" should be supported by a more explicit search of the literature or softened; if prior signed-graph prompt-tuning work exists, it should be cited and discussed.
[§4.4, Eqs. (10)–(11)] It is unclear whether the class prototypes E=[e_P,e_N] and E=[e_{c1},...,e_{cm}] are learnable parameters, computed from the training set, or updated during prompt tuning; please clarify.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: SGPT's components are data-driven constructions and its prompts are tuned on held-out training labels, with no derivation reducing to its own output.

full rationale

The claimed derivation chain is self-contained rather than circular. The graph template constructs multi-hop positive/negative adjacency matrices from the input signed adjacency via Eqs. 4-5, which is a recursive application of balance theory to A+ and A-, not a fitted target. The task template maps node classification and link sign prediction to prototype similarity, and the prompts (feature and semantic) are optimized with loss Eq. 18 on the few-shot training instances and evaluated on unseen test instances. No component is defined in terms of the evaluation target. The balance-theory recursion may be descriptively inaccurate when a node pair is connected by both balanced and unbalanced paths, since the indicator binarizes both channels, but that is a correctness/robustness concern rather than circular reasoning. Citations to the prevalence of balanced structures are external (SGCL [28]) and the balance-theory relation is also supported by external SGCN [3]; the authors' self-citations appear in related work and do not carry the load-bearing argument. The 'first study' claim is a novelty assertion, not a prediction derived from the model. Therefore the paper warrants score 0 on the circularity scale.

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

The central method depends on a handful of hyperparameters (hop number k, basis count r, adapter width d_mid) that are set by sensitivity analysis or left unreported, on the domain assumption that balance theory governs real signed networks, and on the transferability of link-prediction pre-training. No new entities are postulated.

free parameters (3)

k (hop number) = 2
Chosen via sensitivity analysis (Section 5.6); larger k gives marginal gains with exponentially higher construction cost.
r (number of feature prompt basis vectors) = not reported for main results
Varies in sensitivity analysis from 1 to 5 (Figure 4c); default value not stated in the experimental setup.
d_mid (adapter bottleneck dimension) = not reported
Adapter mid-dimension is said to satisfy d_mid << d_out but the exact value is not given.

assumptions (5)

domain assumption Balance theory: a path is positive iff it contains an even number of negative links; balanced paths imply positive relationships, unbalanced paths imply negative relationships.
Used to build positive/negative channels in Section 4.3, Eqs (4)-(5); the paper cites balance theory but does not verify balance levels of the seven datasets.
domain assumption Pre-trained unsigned GNNs aggregate under a homophily assumption, and each separated channel satisfies that assumption.
Motivates the graph template; stated in Section 1 and Section 4.1.
standard math A feature prompt can approximate a graph transformation function (Eq. 12), from GPF.
Assumed from prior work to justify input-space prompting.
domain assumption Unifying downstream tasks into link prediction via class prototypes preserves enough information for node classification and link sign prediction.
Task template in Section 4.3; no analysis of capacity loss.
domain assumption The pre-training link prediction on unsigned graphs produces representations transferable to signed graphs of the same or similar domain.
Central transfer premise; the pre-training graph is not specified in the experimental setup.

how reviews work

0 comments

Cite this review

Pith. "Pith review of SGPT: Few-Shot Prompt Tuning for Signed Graphs." pith.science (2026). https://pith.science/paper/DSGIKHTL

@misc{pith2026241212155,
  author       = {Pith},
  title        = {Pith review of: SGPT: Few-Shot Prompt Tuning for Signed Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DSGIKHTL}},
  note         = {Machine review of arXiv:2412.12155}
}

read the original abstract

Signed Graph Neural Networks (SGNNs) are effective in learning expressive representations for signed graphs but typically require substantial task-specific labels, limiting their applicability in label-scarce industrial scenarios. In contrast, unsigned graph structures are abundant and can be readily leveraged to pre-train Graph Neural Networks (GNNs), offering a promising solution to reduce supervision requirements in downstream signed graph tasks. However, transferring knowledge from unsigned to signed graphs is non-trivial due to the fundamental discrepancies in graph types and task objectives between pre-training and downstream phases. To address this challenge, we propose Signed Graph Prompt Tuning (SGPT), a novel graph prompting framework that adapts pre-trained unsigned GNNs to few-shot signed graph tasks. We first design a graph template based on balance theory to disentangle mixed node relationships introduced by negative links, mitigating the structural mismatches between unsigned and signed graphs. We further introduce a task template that reformulates downstream signed tasks into a unified link prediction objective, aligning their optimization goals with the pre-training task. Furthermore, we develop feature prompts that align downstream semantic spaces with the feature spaces learned during pre-training, and semantic prompts to integrate link sign semantics in a task-aware manner. We conduct extensive experiments on seven benchmark signed graph datasets, demonstrating that SGPT significantly outperforms existing state-of-the-art methods, establishing a powerful and generalizable solution for few-shot signed graph learning.

Figures

Figures reproduced from arXiv: 2412.12155 by the authors.

**Figure 1.** Overall framework of SGPT. The feature prompts are added to the input node features of segregated graph samples before the encoding process to align the downstream semantic space with that of the pre-training task. Afterward, the semantic prompt adaptively integrates the embeddings of graph samples, generating the representations tailored to the requirements of downstream tasks. 4.2 Graph Pre-Training In this pape… view at source ↗

**Figure 2.** Balanced and unbalanced paths in signed graphs. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Impacts of shots on downstream tasks. complex link semantics in signed graphs, demonstrating the necessity of the proposed graph template. (3) As the number of training shots increases, the performance of SDGNN continues to improve. This indicates that supervised SGNNs require adequate training labels to achieve optimal performance. Performance with different GNN backbones. We also assess the flexibility of SGPT wi… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Hyper-parameter analysis results. and signed links, respectively. Let ¯𝑑 = 2𝑙/𝑛 be the average degree, 𝐿 be the number of GNN layers, and 𝑑out and 𝑑mid (𝑑mid ≪𝑑out) denote the output and bottleneck dimensions, respectively. (i) Graph template. A single scan of the edg…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 27 canonical work pages

[1]

Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901

2020
[2]

Dorwin Cartwright and Frank Harary. 1956. Structural balance: a generalization of Heider’s theory. Psychological review 63, 5 (1956), 277

1956
[3]

Tyler Derr, Yao Ma, and Jiliang Tang. 2018. Signed graph convolutional networks. In 2018 IEEE International Conference on Data Mining (ICDM) . IEEE, 929–934

2018
[4]

Taoran Fang, Yunchao Zhang, Yang Yang, Chunping Wang, and Lei Chen. 2024. Universal prompt tuning for graph neural networks. Advances in Neural Infor- mation Processing Systems 36 (2024)

2024
[5]

Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable feature learning for networks. In Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining . 855–864

2016
[6]

Fritz Heider. 1946. Attitudes and cognitive organization.The Journal of psychology 21, 1 (1946), 107–112

1946
[7]

Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. 2019. Parameter-efficient transfer learning for NLP. In International conference on machine learning. PMLR, 2790–2799

2019
[8]

W Hu, B Liu, J Gomes, M Zitnik, P Liang, V Pande, and J Leskovec. 2020. Strategies For Pre-training Graph Neural Networks. In International Conference on Learning Representations (ICLR)

work page 2020

Show all 53 references

[9]

Junjie Huang, Huawei Shen, Liang Hou, and Xueqi Cheng. 2019. Signed graph attention networks. In Artificial Neural Networks and Machine Learning–ICANN 2019: Workshop and Special Sessions: 28th International Conference on Artificial Neu- ral Networks, Munich, Germany, September...

2019
[10]

Junjie Huang, Huawei Shen, Liang Hou, and Xueqi Cheng. 2019. Signed graph attention networks. In International Conference on Artificial Neural Networks . Springer, 566–577

2019
[11]

Junjie Huang, Huawei Shen, Liang Hou, and Xueqi Cheng. 2021. SDGNN: Learn- ing node representation for signed directed networks. In Proceedings of the AAAI conference on artificial intelligence . 196–203

2021
[12]

Srijan Kumar, Francesca Spezzano, and VS Subrahmanian. 2015. Vews: A wikipedia vandal early warning system. In Proceedings of the 21th ACM SIGKDD international conference on knowledge discovery and data mining . 607–616

2015
[13]

Srijan Kumar, Francesca Spezzano, VS Subrahmanian, and Christos Faloutsos
[14]

Jure Leskovec, Daniel Huttenlocher, and Jon Kleinberg. 2010. Signed networks in social media. In Proceedings of the SIGCHI conference on human factors in computing systems. 1361–1370

2010
[15]

Fan Li, Xiaoyang Wang, Dawei Cheng, Cong Chen, Ying Zhang, and Xuemin Lin. 2024. Efficient Dynamic Attributed Graph Generation. arXiv preprint arXiv:2412.08810 (2024)

2024 arXiv
[16]

Fan Li, Zhiyu Xu, Dawei Cheng, and Xiaoyang Wang. 2024. AdaRisk: risk-adaptive deep reinforcement learning for vulnerable nodes detection. IEEE Transactions on Knowledge and Data Engineering 36, 11 (2024), 5576–5590

2024
[17]

Shengrui Li, Xueting Han, and Jing Bai. 2024. AdapterGNN: Parameter-Efficient Fine-Tuning Improves Generalization in GNNs. In Proceedings of the AAAI Con- ference on Artificial Intelligence, Vol. 38. 13600–13608

2024
[18]

Yu Li, Meng Qu, Jian Tang, and Yi Chang. 2023. Signed Laplacian graph neural networks. In Proceedings of the AAAI conference on artificial intelligence . 4444– 4452

2023
[19]

Haoxin Liu, Ziwei Zhang, Peng Cui, Yafeng Zhang, Qiang Cui, Jiashuo Liu, and Wenwu Zhu. 2021. Signed graph neural network with latent groups. In Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining. 1066–1075

2021
[20]

Zemin Liu, Xingtong Yu, Yuan Fang, and Xinming Zhang. 2023. Graphprompt: Unifying pre-training and downstream tasks for graph neural networks. In Pro- ceedings of the ACM Web Conference 2023 . 417–428

2023
[21]

Yihong Ma, Ning Yan, Jiayu Li, Masood Mortazavi, and Nitesh V Chawla. 2024. Hetgpt: Harnessing the power of prompt tuning in pre-trained heterogeneous graph neural networks. In Proceedings of the ACM on Web Conference 2024 . 1015– 1023

2024
[22]

Miller McPherson, Lynn Smith-Lovin, and James M Cook. 2001. Birds of a feather: Homophily in social networks. Annual review of sociology 27, 1 (2001), 415–444

2001
[24]

Pedro Mercado, Jessica Bosch, and Martin Stoll. 2020. Node classification for signed social networks using diffuse interface methods. In Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2019, Würzburg, Germany, September 16–20, 2019, Procee...

2020
[25]

Mathias Niepert, Mohamed Ahmed, and Konstantin Kutzkov. 2016. Learning convolutional neural networks for graphs. In International conference on machine learning. PMLR, 2014–2023

2016
[26]

Bryan Perozzi, Rami Al-Rfou, and Steven Skiena. 2014. Deepwalk: Online learning of social representations. In Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining . 701–710

2014
[27]

Zhengxiang Shi and Aldo Lipani. 2023. Don’t stop pretraining? make prompt- based fine-tuning powerful learner. Advances in Neural Information Processing Systems 36 (2023), 5827–5849

2023
[28]

Lin Shu, Erxin Du, Yaomin Chang, Chuan Chen, Zibin Zheng, Xingxing Xing, and Shaofeng Shen. 2021. Sgcl: Contrastive representation learning for signed graphs. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management. 1671–1680

2021
[29]

Mingchen Sun, Kaixiong Zhou, Xin He, Ying Wang, and Xin Wang. 2022. Gppt: Graph pre-training and prompt tuning to generalize graph neural networks. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 1717–1727

2022
[30]

Renjie Sun, Chen Chen, Xiaoyang Wang, Ying Zhang, and Xun Wang. 2022. Stable Community Detection in Signed Social Networks. IEEE Trans. Knowl. Data Eng. 34, 10 (2022), 5051–5055

2022
[31]

Renjie Sun, Yanping Wu, Xiaoyang Wang, Chen Chen, Wenjie Zhang, and Xuemin Lin. 2024. Efficient Balanced Signed Biclique Search in Signed Bipartite Graphs. TKDE 36, 3 (2024), 1069–1083

2024
[32]

Xiangguo Sun, Hong Cheng, Jia Li, Bo Liu, and Jihong Guan. 2023. All in one: Multi-task prompting for graph neural networks. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . 2120–2131

2023
[33]

Xingyu Tan, Jingya Qian, Chen Chen, Sima Qing, Yanping Wu, Xiaoyang Wang, and Wenjie Zhang. 2023. Higher-order peak decomposition. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management

2023
[34]

Xingyu Tan, Xiaoyang Wang, Qing Liu, Xiwei Xu, Xin Yuan, and Wenjie Zhang
[35]

Xingyu Tan, Xiaoyang Wang, Qing Liu, Xiwei Xu, Xin Yuan, Liming Zhu, and Wenjie Zhang. 2025. Hydra: Structured Cross-Source Enhanced Large Language Model Reasoning. arXiv:2505.17464 [cs.CL] https://arxiv.org/abs/2505.17464

2025
[36]

Jiliang Tang, Charu Aggarwal, and Huan Liu. 2016. Node classification in signed social networks. In Proceedings of the 2016 SIAM international conference on data mining. SIAM, 54–62

2016
[37]

Jiliang Tang, Yi Chang, Charu Aggarwal, and Huan Liu. 2016. A survey of signed network mining in social media. ACM Computing Surveys (CSUR) 49, 3 (2016), 1–37

2016
[38]

Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Liò, and Yoshua Bengio. 2018. Graph Attention Networks. In International Con- ference on Learning Representations

2018
[39]

Jinghao Wang, Yanping Wu, Xiaoyang Wang, Chen Chen, Ying Zhang, and Lu Qin. 2025. Effective Influence Maximization with Priority. In Proceedings of the ACM on Web Conference 2025. 4673–4683

2025
[40]

Jinghao Wang, Yanping Wu, Xiaoyang Wang, Ying Zhang, Lu Qin, Wenjie Zhang, and Xuemin Lin. 2024. Efficient influence minimization via node blocking. arXiv preprint arXiv:2405.12871 (2024)

2024 arXiv
[41]

Junyang Wu, Xianhang Li, Chen Wei, Huiyu Wang, Alan Yuille, Yuyin Zhou, and Cihang Xie. 2022. Unleashing the power of visual prompting at the pixel level. arXiv preprint arXiv:2212.10556 (2022)

2022 arXiv
[42]

Xuansheng Wu, Kaixiong Zhou, Mingchen Sun, Xin Wang, and Ninghao Liu. 2023. A survey of graph prompting methods: techniques, applications, and challenges. arXiv preprint arXiv:2303.07275 (2023)

2023 arXiv
[43]

Yanping Wu, Renjie Sun, Chen Chen, Xiaoyang Wang, and Qiuyu Zhu. 2020. Maximum signed (k, r)-truss identification in signed networks. In Proceedings of the 29th ACM International Conference on Information & Knowledge Management . 3337–3340

2020
[44]

Yuning You, Tianlong Chen, Yang Shen, and Zhangyang Wang. 2021. Graph contrastive learning automated. In International Conference on Machine Learning . PMLR, 12121–12132

2021
[45]

Xingtong Yu, Yuan Fang, Zemin Liu, and Xinming Zhang. 2024. Hgprompt: Bridging homogeneous and heterogeneous graphs for few-shot prompt learning. In Proceedings of the AAAI Conference on Artificial Intelligence . 16578–16586. SGPT: Few-Shot Prompt Tuning for Signed Graphs CIKM...

2024
[47]

Xingtong Yu, Zhenghao Liu, Yuan Fang, Zemin Liu, Sihong Chen, and Xinming Zhang. 2024. Generalized graph prompt: Toward a unification of pre-training and downstream tasks on graphs. IEEE Transactions on Knowledge and Data Engineering (2024)

2024
[48]

Muhan Zhang and Yixin Chen. 2018. Link prediction based on graph neural networks. Advances in neural information processing systems 31 (2018)

2018
[49]

Yunrui Zhang, Gustavo Batista, and Salil S Kanhere. 2025. Instance-Wise Mono- tonic Calibration by Constrained Transformation. arXiv preprint arXiv:2507.06516 (2025)

2025 arXiv
[50]

Yunrui Zhang, Gustavo Batista, and Salil S Kanhere. 2025. Label Shift Estimation With Incremental Prior Update. In Proceedings of the 2025 SIAM International Conference on Data Mining (SDM) . SIAM, 134–142

2025
[51]

Zeyu Zhang, Jiamou Liu, Kaiqi Zhao, Song Yang, Xianda Zheng, and Yifei Wang
[52]

Zeyu Zhang, Peiyao Zhao, Xin Li, Jiamou Liu, Xinrui Zhang, Junjie Huang, and Xiaofeng Zhu. 2024. Signed Graph Representation Learning: A Survey. arXiv preprint arXiv:2402.15980 (2024)

2024 arXiv
[2016]

In 2016 IEEE 16th International Conference on Data Mining (ICDM)

Edge weight prediction in weighted signed networks. In 2016 IEEE 16th International Conference on Data Mining (ICDM) . IEEE, 221–230

2016
[2023]

In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval

Contrastive learning for signed bipartite graphs. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1629–1638
[2025]

In Proceedings of the ACM on Web Conference 2025

Paths-over-graph: Knowledge graph empowered large language model reasoning. In Proceedings of the ACM on Web Conference 2025 . 3505–3522

2025

Pith tools

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

[1] [1]

Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901

2020

[2] [2]

Dorwin Cartwright and Frank Harary. 1956. Structural balance: a generalization of Heider’s theory. Psychological review 63, 5 (1956), 277

1956

[3] [3]

Tyler Derr, Yao Ma, and Jiliang Tang. 2018. Signed graph convolutional networks. In 2018 IEEE International Conference on Data Mining (ICDM) . IEEE, 929–934

2018

[4] [4]

Taoran Fang, Yunchao Zhang, Yang Yang, Chunping Wang, and Lei Chen. 2024. Universal prompt tuning for graph neural networks. Advances in Neural Infor- mation Processing Systems 36 (2024)

2024

[5] [5]

Aditya Grover and Jure Leskovec. 2016. node2vec: Scalable feature learning for networks. In Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining . 855–864

2016

[6] [6]

Fritz Heider. 1946. Attitudes and cognitive organization.The Journal of psychology 21, 1 (1946), 107–112

1946

[7] [7]

Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. 2019. Parameter-efficient transfer learning for NLP. In International conference on machine learning. PMLR, 2790–2799

2019

[8] [8]

W Hu, B Liu, J Gomes, M Zitnik, P Liang, V Pande, and J Leskovec. 2020. Strategies For Pre-training Graph Neural Networks. In International Conference on Learning Representations (ICLR)

work page 2020