Pith. sign in

REVIEW 4 major objections 6 minor 42 references

TwiUSD: A Benchmark Dataset and Structure-Aware LLM Framework for User Stance Detection

T0 review · 4 major / 6 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 A new benchmark and model show that user-level stance detection gains from filtering followee tweets and routing features by graph structure.

desk verdict A useful new user-level stance dataset with a fixable overclaim; the follower-label circularity needs a reannotation check. read the letter →

arxiv 2506.13343 v2 pith:GGP3XMTG submitted 2025-06-16 cs.SI

Fuqiang Niu , Zini Chen , Zhiyu Xie , Hu Huang , Genan Dai , Bowen Zhang This is my paper · ORCID
classification cs.SI
keywords user-levelstancedetectionTwiUSDbenchmarkfolloweerelationshipssocialgraphLLMrelevancefilteringfeatureroutingneuralnetworks2020U.S.election
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

User-level stance detection—deciding a Twitter user's overall position on a target such as a presidential candidate—has been held back by noisy labels produced from hashtags or retweets and by datasets that ignore social links. This paper introduces TwiUSD, which it describes as the first large-scale manually annotated user-level stance benchmark with explicit followee relationships: 16,211 users and 47,757 tweets about the 2020 U.S. election, labeled toward Joe Biden and Donald Trump. On that resource the paper builds MRFG, a framework that first lets a large language model discard followee tweets irrelevant to the user, then splits the user's feature dimensions by a graph-informativeness score and routes the structure-sensitive ones through a relational graph network and the rest through an MLP. The paper reports that MRFG clearly beats content-only pre-trained models and graph baselines in in-target tests—Biden Favg 84.19 versus RoBERTa's 79.21—and stays competitive when training and test targets are swapped. If the benchmark and method hold up, the field gains a realistic testbed that joins language with social structure instead of relying on hashtag heuristics.

What carries the argument

The load-bearing machinery is the pairing of the user-followee graph with S-TFI feature ranking. For user features $X$ and normalized adjacency matrix $\hat{A}$, the informativeness of dimension $m$ is $TFI_m = I(Y; \tilde{X}_{:,m})$ with $\tilde{X} = \hat{A}X$, meaning the mutual information between that feature dimension and stance labels after one step of graph propagation. Features with high TFI are treated as structure-sensitive and processed by a relational graph convolutional network; low-TFI features are treated as content-sensitive and processed by an MLP, and the two encoded halves are concatenated before classification. Before feature routing, a large language model scores each followee tweet against the user's own content and deletes tweets scored as irrelevant, so the social context entering the model is selected rather than raw. This design is what lets the model choose context and delegate features to the encoder that can use them.

What would settle it

Reannotate a random sample of follower users from their own tweets and profiles alone, with annotators blind to followee stances, and measure agreement with TwiUSD's labels; if agreement is far below the reported kappa values of 0.90 and 0.91, the follower labels are substantially determined by followee stances. A complementary check: train MRFG without any followee tweets as input and compare follower accuracy; a large drop would indicate the model is copying followee stance rather than detecting an independent user stance.

Watch

Extended reading notes

Core claim

The paper's central claim has two parts. First, TwiUSD supplies a user-level stance benchmark whose labels are assigned not only from a user's own tweets and profile but, for the 73.29% of users who follow others, from the stances of the users they follow; this is meant to mirror stance propagation in real social networks and to make structure usable as supervision. Second, MRFG outperforms strong baselines by treating social context as noisy rather than aggregating all neighbors: an LLM filtering unit scores each followee tweet as 1, 2, or 3 for relevance to the target user, a structural topological feature informativeness (S-TFI) score ranks each embedding dimension by the mutual information it carries about labels after graph smoothing, and the top-ranked dimensions go through a relational graph convolutional network while the rest go through an MLP. The headline empirical results are in-target Favg of 84.19 on Biden and 81.27 on Trump, above RoBERTa's 79.21 and 76.34, with the paper reporting statistically significant gains over its strongest baselines. In cross-target evaluation the paper's model is competitive, and the paper notes a trade-off between target-specific feature selection and transfer.

Load-bearing premise

The load-bearing premise is the annotation rule that a follower user's gold label is inferred by jointly considering the follower's own content and the stances of the users they follow; if that rule does not capture a stance the user would hold independently, nearly three-quarters of the benchmark's labels partly encode the annotation rule itself.

Editorial extensions

If this is right

  • Manual, followee-aware annotation can replace noisy hashtag heuristics: applying two published hashtag labeling rules to TwiUSD mislabels more than half of users, so the benchmark raises the evaluation bar.
  • LLM-based relevance filtering pays off: removing the filter drops Biden Favg from 84.19 to 81.33, and cosine-similarity filtering is consistently worse.
  • Feature routing is the active ingredient: forcing all features through either the RGCN or the MLP alone reduces performance, so splitting by graph-informativeness is what carries the gain.
  • The benchmark's 70/15/15 splits for two targets make both in-target and cross-target evaluation reproducible, letting models be tested for transfer between candidates.
  • Because follower labels are inferred jointly with followee stances, any model trained on TwiUSD is being rewarded for using social propagation signals, not just text.

Reading between the lines

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

  • A content-only reannotation of follower users would settle whether TwiUSD measures stance a user holds independently or stance inferred from social influence; the paper does not report such a check.
  • The paper's own cross-target numbers suggest the TFI ranking, computed on training-target labels, is the main obstacle to transfer; reranking features on the target side or learning a transferable routing policy is a natural next step.
  • The filter-and-route recipe should transfer to other user-level social tasks, such as bot detection or preference inference, whenever noisy neighborhood text accompanies a user graph.
  • Error analysis singles out single-tweet users and users with conflicting followees as the failure frontier, so increasing tweets per user or adding a conflict-resolution step would likely raise the ceiling.
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

4 major / 6 minor

Summary. The paper introduces TwiUSD, a user-level stance detection benchmark built from TwiBot-22 and filtered for the 2020 U.S. presidential election, with 16,211 users, 47,757 tweets, explicit follow relationships, and favor/against/none labels obtained through a manual annotation protocol. It also proposes MRFG, a structure-aware framework that uses an LLM-based relevance filter to select followee tweets and a TFI-based feature splitting mechanism to route graph-favored features through an RGCN and graph-disfavored features through an MLP. Experiments report strong in-target results for MRFG over PLMs, graph-based models, and LLM prompting baselines, as well as ablation studies, sensitivity analysis over the feature selection ratio r, and an error analysis.

Significance. If the in-target results hold, TwiUSD would be a valuable resource for user-level stance detection because it is large, manually annotated, and explicitly incorporates follow relationships, with high inter-annotator agreement (kappa 0.90 for Biden and 0.91 for Trump). The MRFG framework is also a sensible engineering contribution: the ablation studies isolating LLM-FU and the S-TFI-based dual-path design, the comparison against cosine-similarity filtering, and the error analysis are useful and generally credible. However, the benchmark's annotation rule for follower users, the overstatement of the cross-target results, and the absence of variance reporting currently prevent acceptance. The code and dataset are not yet released, so the reproducibility claims cannot be independently verified.

major comments (4)
  1. [Abstract, Section 6.2, Table 4, Conclusion] The abstract and conclusion claim that MRFG 'consistently outperforms strong baselines' in both in-target and cross-target evaluation, but Table 4 contradicts this. In Trump→Biden, MRFG achieves Favg 45.16 and Acc 43.24, below BERT (47.49, 49.13) and JointCL (48.38, 44.86); in Biden→Trump, MRFG's Favg 46.08 is below BERT-GCN's 49.03. Section 6.2 itself concedes that 'BERT and JoinCL achieve the best Favg or accuracy in at least one direction' and that TFI-based feature selection 'may not transfer well.' The headline empirical claim must be revised to state that MRFG is best in in-target evaluation and competitive, but not universally best, in cross-target evaluation.
  2. [Section 3.3, Figure 1, Eq. (1)] The benchmark construction has a circularity problem that bears directly on the validity of the reported in-target gains. According to Section 3.3, labels for follower users (73.29% of all users) are 'inferred by jointly considering their own content and the stances of their followees,' while MRFG receives followee tweets TF(u) as model input in Eq. (1) and propagates over the follow graph. A model can therefore achieve high accuracy on the follower portion by learning to reproduce the annotation rule rather than by independently detecting the user's stance. The paper provides no check that follower labels agree with a purely content-based reannotation of the same users, and no breakdown of performance by user category. At minimum, the authors should report results separately for followees, followers, and isolated users, and provide evidence that the benchmark measures stance beyond annotation-rule replication.
  3. [Section 6, Tables 3 and 4] The paper states that 'the reported results are averaged over three runs with different random seeds' but reports no standard deviations, confidence intervals, or significance tests for most comparisons. The paired t-test in Section 6.1 is only reported against JoinCL and TPDG, not against the strongest baselines (BERT and RoBERTa) or against cross-target baselines. Because the cross-target gaps are small and the central claim depends on point estimates, the authors should report variance and perform significance tests for the main in-target and cross-target comparisons.
  4. [Table 1, Table 2, Section 3.4] The dataset size is internally inconsistent. Table 1 reports Stage-2 counts of 8,348 users for Biden and 10,837 users for Trump, which sum to 19,185 users, while Section 3.4 and Table 2 state a total of 16,211 users. The paper does not explain whether users overlap across the two target subsets or how such overlap is handled in the 70/15/15 splits. The authors should clarify whether a user can appear in both target subsets and, if so, how train/test leakage across targets is prevented.
minor comments (6)
  1. [Section 3.3] The computation of Cohen's kappa is described as using only the 'Favor' and 'Against' classes, following Li et al. (2021), but the task has three classes including 'None.' The paper should clarify how 'None' instances are treated in the agreement calculation.
  2. [Figure 3] The axis labels in Figure 3 appear as unicode escape artifacts (e.g., '/uni00000013/uni00000011/...'), making the figure unreadable; the figure needs to be regenerated.
  3. [Section 3.1] The paper says bot accounts were filtered out but does not describe the filtering procedure or its accuracy; since the benchmark's credibility depends on using human users only, a brief description of the bot classifier or criteria should be added.
  4. [Section 6.5] The sensitivity analysis in Figure 3 suggests that performance peaks at r in [0.3, 0.4], but Tables 3 and 4 only report results for r = 0.3. The authors should state whether r = 0.3 was chosen before seeing the test set or report results across the full range with variance.
  5. [Ethics Statement] There is a typo, 'termss,' and the disclosure that ChatGPT was used for writing is not related to the scientific content; the sentence should be corrected or removed.
  6. [Section 3.4] The claim that TwiUSD 'surpasses the diversity of most existing user-level stance datasets' is vague and unsupported; a concrete comparison of user counts, tweet counts, and label distributions with prior datasets would be more informative.

Circularity Check

1 steps flagged · score 5.0 of 10

In-target superiority is partially circular: follower labels are inferred from followee stances, and MRFG is given followee tweets as input.

  1. self definitional [Section 3.3 (annotation protocol), Section 3.4 (73.29% followers), Section 4.1 Eq. 1 (model input)]
    ""Then, for follower users, stance labels are inferred by jointly considering their own content and the stances of their followees, mirroring realistic stance propagation patterns in social networks." "As shown in Figure 1, the majority of users are followers (73.29%)." "Each user is associated with two types of content: their own tweet collection Tu and the tweet set of users they follow (i.e., followees) TF(u)... f(u) = arg max P(y | Tu, TF(u), du, Ga).""

    For 73.29% of users, the gold label was generated by an annotation rule that explicitly uses the stances of followees. Eq. 1 gives the model the followees' tweets TF(u) as a direct input, and MRFG's LLM-FU is designed to keep followee tweets relevant to the target user. A model can therefore achieve high in-target accuracy by reproducing the annotation rule (e.g., following the dominant or relevant followee stance) rather than detecting an independently observable stance. The paper provides no re-annotation of follower users from their own content alone and no separate evaluation for followers versus followees or isolated users, so the reported in-target advantage on the follower majority cannot be separated from rule replication.

full rationale

The model equations themselves are not circular: MRFG is a standard feature-routing RGCN/MLP pipeline trained with cross-entropy, and no parameter is fitted to the test set. The load-bearing circularity is in benchmark construction: follower gold labels for 73.29% of users are inferred jointly from followee stances, and the model receives followee tweets as an input feature. That makes the in-target evaluation partly a replication of the annotation rule rather than an independent measure of user stance. However, the paper's own content-only baselines (BERT, RoBERTa) are far below MRFG and do not use followee tweets in the same way, so the result is not forced by construction; a substantial independent component remains. The cross-target claim in the abstract is contradicted by Table 4 and Section 6.2, but that is a correctness and verifiability failure, not circularity. No load-bearing self-citation chain or imported uniqueness theorem was found, and citations to the authors' prior survey and positioning work are not load-bearing. Score 5 reflects one partial but substantial construction-level circularity.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The dataset and framework rest on several unproven domain assumptions. Most load-bearing is that follower stance can be derived from followee stance, since that defines 73% of the gold labels. The hashtag set is asserted neutral despite partisan entries. GPT-4o's relevance scores are taken as reliable. The only tuned scalar is the feature-split ratio r=0.3. No invented physical or formal entities are introduced.

free parameters (3)
  • Feature selection ratio r = 0.3
    Controls the split between RGCN-processed and MLP-processed feature dimensions; selected from the range that peaks at 0.3-0.4 in Figure 3 on both targets. This is a dataset-tuned hyperparameter.
  • LLM relevance cutoff = retain scores >= 2, discard score 1
    The GPT-4o filter boundary is set by hand; the paper does not evaluate alternative cutoffs or validate the LLM scores against human relevance judgments.
  • Hashtag filtering set = 16 hashtags including #Trump2020LandSlide, #BidenForPresident, #SleepyJoe
    The hand-chosen keyword set determines which users and tweets enter TwiUSD. Despite being described as ideologically neutral, several tags are openly partisan and shape the label distribution.
assumptions (6)
  • domain assumption Follow relationships are persistent, meaningful ties for modeling social influence.
    Section 3.2 builds the user graph on follow edges and densifies it using followees' followers, assuming these edges carry stance-relevant signal.
  • domain assumption Follower stance can be obtained by combining the user's own content with followee stances.
    Section 3.3 defines the annotation protocol for 73.29% of users. This is the main load-bearing assumption behind the dataset labels.
  • domain assumption The selected hashtags are politically salient yet ideologically neutral and jointly identify target-relevant tweets.
    Appendix B lists tags like #Trump2020LandSlide and #BidenForPresident, which are target-supporting, so neutrality is not established; this affects topic coverage and label distribution.
  • domain assumption GPT-4o relevance scores in {1,2,3} are sufficiently accurate for filtering followee tweets.
    Section 4.3(1) and Appendix D use LLM outputs as ground truth for relevance without human validation or confidence calibration.
  • domain assumption S-TFI feature informativeness, as defined in Eq. (2), ranks feature dimensions correctly for the graph-splitting decision.
    Section 4.3(2) adopts TFI from Zheng et al. (2025); if TFI misranks dimensions, the RGCN/MLP routing has no justification. TFI is computed from training labels only, which limits leakage.
  • standard math Mutual information I(Y; X_tilde) can be reliably estimated from the discretized training features.
    TFI requires MI estimation on smoothed BERT features; the paper does not specify the estimator or discretization, so it treats the estimate as trustworthy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TwiUSD: A Benchmark Dataset and Structure-Aware LLM Framework for User Stance Detection." pith.science (2026). https://pith.science/paper/GGP3XMTG

@misc{pith2026250613343,
  author       = {Pith},
  title        = {Pith review of: TwiUSD: A Benchmark Dataset and Structure-Aware LLM Framework for User Stance Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GGP3XMTG}},
  note         = {Machine review of arXiv:2506.13343}
}
read the original abstract

User-level stance detection (UserSD) remains challenging due to the lack of high-quality benchmarks that jointly capture linguistic and social structure. In this paper, we introduce TwiUSD, the first large-scale, manually annotated UserSD benchmark with explicit followee relationships, containing 16,211 users and 47,757 tweets. TwiUSD enables rigorous evaluation of stance models by integrating tweet content and social links, with superior scale and annotation quality. Building on this resource, we propose MRFG: a structure-aware framework that uses LLM-based relevance filtering and feature routing to address noise and context heterogeneity. MRFG employs multi-scale filtering and adaptively routes features through graph neural networks or multi-layer perceptrons based on topological informativeness. Experiments show MRFG consistently outperforms strong baselines (including PLMs, graph-based models, and LLM prompting) in both in-target and cross-target evaluation.

Figures

Figures reproduced from arXiv: 2506.13343 by the authors.

Figure 1
Figure 1. User and tweet distributions by category in [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The architecture of our MRFG framework. on structure-aware feature selection (Zheng et al., 2025), MRF further assesses which features are most suitable for graph-based learning, thereby enabling more effective stance reasoning. MRF operates in two main steps: (1) Relevance Filtering via LLM-FU. We first employ a LLM-based Filtering Unit (LLM-FU) to evaluate the semantic relevance between each fol￾lowee tweet TF(u) … view at source ↗
Figure 3
Figure 3. Comparison of LLM-based vs. Cosine-based [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 23 canonical work pages

  1. [1]

    Abeer AlDayel and Walid Magdy. 2021. Stance detection on social media: State of the art and trends. Information Processing & Management, 58(4):102597

    2021

  2. [2]

    Zefan Cai, Baobao Chang, and Wenjuan Han. 2023. Human-in-the-loop through chain-of-thought. arXiv preprint arXiv:2306.07932

    arXiv 2023

  3. [3]

    Aupetit, and Preslav Nakov

    Kareem Darwish, Peter Stefanov, Micha \" e l J. Aupetit, and Preslav Nakov. 2020. Unsupervised user stance detection on twitter. In Proceedings of the Fourteenth International AAAI Conference on Web and Social Media , pages 141--152

    work page 2020

  4. [4]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT : Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) , pages 4171--4186

    2019

  5. [5]

    Kuntal Dey, Ritvik Shrivastava, and Saroj Kaushik. 2018. Topical stance detection for twitter: A two-phase lstm model using attention. In European Conference on Information Retrieval, pages 529--536. Springer

    work page 2018

  6. [6]

    Jiachen Du, Ruifeng Xu, Yulan He, and Lin Gui. 2017. Stance classification with target-specific neural attention. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence, IJCAI-17 , pages 3988--3994

    work page 2017

  7. [7]

    Elzanfaly, Zeyad Radwan, and Nermin Abdelhakim Othman

    Doaa S. Elzanfaly, Zeyad Radwan, and Nermin Abdelhakim Othman. 2023. User stance detection and prediction considering most frequent interactions. In Artificial Intelligence and Online Engineering, pages 421--433, Cham. Springer International Publishing

    work page 2023

  8. [8]

    Shangbin Feng, Zhaoxuan Tan, Herun Wan, Ningnan Wang, Zilong Chen, Binchi Zhang, Qinghua Zheng, Wenqian Zhang, Zhenyu Lei, Shujie Yang, and 1 others. 2022. Twibot-22: Towards graph-based twitter bot detection. Advances in Neural Information Processing Systems, 35:35254--35269

    work page 2022

Show all 42 references
  1. [9]

    Margherita Gambini, Caterina Senette, Tiziano Fagni, and Maurizio Tesconi. 2023. From tweets to stance: An unsupervised framework for user stance detection on twitter. In Discovery Science, pages 96--110, Cham. Springer Nature Switzerland

    2023

  2. [10]

    Joseph Gatto, Omar Sharif, and Sarah Preum. 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.273 Chain-of-thought embeddings for stance detection on social media . In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 4154--4161, Singapore. Assoc...

    2023 doi

  3. [11]

    Kyle Glandt, Sarthak Khanal, Yingjie Li, Doina Caragea, and Cornelia Caragea. 2021. Stance detection in covid-19 tweets. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language P...

    2021

  4. [12]

    Hu Huang, Bowen Zhang, Yangyang Li, Baoquan Zhang, Yuxi Sun, Chuyao Luo, and Cheng Peng. 2023. Knowledge-enhanced prompt-tuning for stance detection. ACM Transactions on Asian and Low-Resource Language Information Processing, 22(6):1--20

    2023

  5. [13]

    Kornraphop Kawintiranon and Lisa Singh. 2021. https://doi.org/10.18653/v1/2021.naacl-main.376 Knowledge enhanced masked language model for stance detection . In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: H...

    2021 doi

  6. [14]

    Dilek K \"u c \"u k and Fazli Can. 2020. Stance detection: A survey. ACM Computing Surveys (CSUR), 53(1):1--37

    2020

  7. [15]

    Xiaochong Lan, Chen Gao, Depeng Jin, and Yong Li. 2024. Stance detection with collaborative role-infused llm-based agents. In Proceedings of the international AAAI conference on web and social media, volume 18, pages 891--903

    2024

  8. [16]

    Yu, and Lifang He

    Chen Li, Hao Peng, Jianxin Li, Lichao Sun, Lingjuan Lyu, Lihong Wang, Philip S. Yu, and Lifang He. 2022. Joint stance and rumor detection in hierarchical heterogeneous graph. IEEE Trans. Neural Networks Learn. Syst. , 33(6):2530--2542

    2022

  9. [17]

    Yingjie Li, Tiberiu Sosea, Aditya Sawant, Ajith Jayaraman Nair, Diana Inkpen, and Cornelia Caragea. 2021. P-stance: A large dataset for stance detection in political domain. In Findings of the Association for Computational Linguistics: ACL/IJCNLP Online Event, August 1-6

    2021

  10. [18]

    Yupeng Li, Haorui He, Shaonan Wang, Francis CM Lau, and Yunya Song. 2023 a . Improved target-specific stance detection on social media platforms by delving into conversation threads. IEEE Transactions on Computational Social Systems

    2023

  11. [19]

    Yupeng Li, Dacheng Wen, Haorui He, Jianxiong Guo, Xuan Ning, and Francis CM Lau. 2023 b . Contextual target-specific stance detection on twitter: Dataset and method. In 2023 IEEE International Conference on Data Mining (ICDM), pages 359--367. IEEE

    2023

  12. [20]

    Bin Liang, Yonghao Fu, Lin Gui, Min Yang, Jiachen Du, Yulan He, and Ruifeng Xu. 2021. Target-adaptive graph for cross-target stance detection. In WWW '21: The Web Conference 2021, Virtual Event / Ljubljana, Slovenia, April 19-23 , pages 3453--3464

    2021

  13. [21]

    Bin Liang, Ang Li, Jingqian Zhao, Lin Gui, Min Yang, Yue Yu, Kam-Fai Wong, and Ruifeng Xu. 2024. https://doi.org/10.18653/v1/2024.findings-acl.736 Multi-modal stance detection: New datasets and model . In Findings of the Association for Computational Linguistics: ACL 2024, pag...

    2024 doi

  14. [22]

    Bin Liang, Qinlin Zhu, Xiang Li, Min Yang, Lin Gui, Yulan He, and Ruifeng Xu. 2022. Jointcl: a joint contrastive learning framework for zero-shot stance detection. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers...

    2022

  15. [23]

    Rui Liu, Zheng Lin, Yutong Tan, and Weiping Wang. 2021. Enhancing zero-shot and few-shot stance detection with commonsense knowledge graph. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3152--3157

    2021

  16. [24]

    Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. https://arxiv.org/abs/1907.11692 Roberta: A robustly optimized BERT pretraining approach . CoRR, abs/1907.11692

    2019 arXiv

  17. [25]

    Mary L McHugh. 2012. Interrater reliability: the kappa statistic. Biochemia medica, 22(3):276--282

    2012

  18. [26]

    Saif Mohammad, Svetlana Kiritchenko, Parinaz Sobhani, Xiao - Dan Zhu, and Colin Cherry. 2016. Semeval-2016 task 6: Detecting stance in tweets. In Proceedings of the 10th International Workshop on Semantic Evaluation, SemEval NAACL-HLT, San Diego, CA, USA, June 16-17, pages 31--41

    2016

  19. [27]

    Mohammad, Parinaz Sobhani, and Svetlana Kiritchenko

    Saif M. Mohammad, Parinaz Sobhani, and Svetlana Kiritchenko. 2017. Stance and sentiment in tweets. ACM Trans. Internet Technol., 17(3)

    2017

  20. [28]

    Fuqiang Niu, Min Yang, Ang Li, Baoquan Zhang, Xiaojiang Peng, and Bowen Zhang. 2024. A challenge dataset and effective models for conversational stance detection. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Eva...

    2024

  21. [29]

    Younes Samih and Kareem Darwish. 2021. https://doi.org/10.18653/v1/2021.eacl-main.227 A few topical tweets are enough for effective user stance detection . In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volu...

    2021 doi

  22. [30]

    Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, and Max Welling

    Michael Schlichtkrull, Thomas N. Kipf, Peter Bloem, Rianne van den Berg, Ivan Titov, and Max Welling. 2017. https://arxiv.org/abs/1703.06103 Modeling relational data with graph convolutional networks . Preprint, arXiv:1703.06103

    2017 arXiv

  23. [31]

    Taylor Shin, Yasaman Razeghi, Robert L Logan IV, Eric Wallace, and Sameer Singh. 2020. Autoprompt: Eliciting knowledge from language models with automatically generated prompts. arXiv preprint arXiv:2010.15980

    2020 arXiv

  24. [32]

    Yuanfu Sun, Zhengnan Ma, Yi Fang, Jing Ma, and Qiaoyu Tan. 2025. https://aclanthology.org/2025.findings-naacl.131/ G raph ICL : Unlocking graph learning potential in LLM s through structured prompt design . In Findings of the Association for Computational Linguistics: NAACL 20...

    2025

  25. [33]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. 2023. https://arxiv.org/abs/2302.13971 Llama:...

    2023 arXiv

  26. [34]

    Ramon Villa-Cox, Sumeet Kumar, Matthew Babcock, and Kathleen M Carley. 2020. Stance in replies and quotes (srq): A new dataset for learning stance in twitter conversations. arXiv preprint arXiv:2006.00691

    2020 arXiv

  27. [35]

    Chang Xu, Cecile Paris, Surya Nepal, and Ross Sparks. 2018. Cross-target stance classification with self-attention networks. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 778--783

    2018

  28. [36]

    Bowen Zhang, Genan Dai, Fuqiang Niu, Nan Yin, Xiaomao Fan, Senzhang Wang, Xiaochun Cao, and Hu Huang. 2024 a . A survey of stance detection on social media: New directions and perspectives. arXiv preprint arXiv:2409.15690

    2024 arXiv

  29. [37]

    Bowen Zhang, Xianghua Fu, Daijun Ding, Hu Huang, Yangyang Li, and Liwen Jing. 2023. Investigating chain-of-thought with chatgpt for stance detection on social media. arXiv preprint arXiv:2304.03087

    2023 arXiv

  30. [38]

    Chong Zhang, Zhenkun Zhou, Xingyu Peng, and Ke Xu. 2024 b . Doubleh: Twitter user stance detection via bipartite graph neural networks. In Proceedings of the Eighteenth International AAAI Conference on Web and Social Media , pages 1766--1778

    2024

  31. [39]

    Yilun Zheng, Xiang Li, Sitao Luan, Xiaojiang Peng, and Lihui Chen. 2025. https://openreview.net/forum?id=I9omfcWfMp Let your features tell the differences: Understanding graph convolution by feature splitting . In The Thirteenth International Conference on Learning Representations

    2025

  32. [40]

    Lixing Zhu, Yulan He, and Deyu Zhou. 2020. https://doi.org/10.1016/j.ipm.2019.03.010 Neural opinion dynamics model for the prediction of user-level stance dynamics . Information Processing & Management, 57(2):102031

    2020 doi

  33. [41]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  34. [42]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

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