Pith. sign in

REVIEW 3 major objections 7 minor 43 references

Unsupervised Anomaly Detection of Information Operations Users via Behavioral and Language Patterns

T0 review · 3 major / 7 minor · reviewed 2026-07-08 · glm-5.2

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

Pith's one-line read Fusing posting-time patterns with LLM judgments detects coordinated influence operations

desk verdict TENSOR fuses TPP behavioral anomaly scores with LLM-derived evidence via EPHAD for unsupervised IO detection; the fusion works but the contamination-mitigation framing is not supported by the data. read the letter →

arxiv 2607.05855 v1 pith:WTYNF63W submitted 2026-07-07 cs.LG cs.AI

Sishun Liu , Sajal Halder , Ke Deng , Yan Wang , Xiuzhen Zhang This is my paper · ORCID
classification cs.LGcs.AI
keywords detectionuseruserstemporalbehavioralexistinginformationpatterns
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 proposes TENSOR, a method for detecting Information Operations (IO) users on social media without requiring labeled training data. The core idea is that IO users are rare and exhibit distinct temporal behavioral patterns (e.g., coordinated posting times) that differ from normal users. TENSOR trains a Temporal Point Process (TPP) on user posting timestamps to flag anomalous behavioral patterns, but this TPP is trained on contaminated data containing both normal and IO users, which degrades its accuracy. To correct for this, the paper introduces an evidence function: a Large Language Model (LLM) is prompted with each user's post timeline and its text output is converted into a quantitative score via semantic similarity mapping. This LLM-derived score is then fused with the TPP score through exponential tilting (EPHAD), a post-hoc adjustment mechanism that corrects the TPP's predictions at test time. The paper demonstrates that neither the TPP alone nor the LLM alone performs well, but their fusion yields a system that outperforms existing unsupervised baselines across five real-world IO datasets.

What carries the argument

Temporal Point Process (TPP), exponential tilting (EPHAD), evidence function from LLM semantic similarity

What would settle it

Construct a dataset where the LLM's IO judgments are systematically biased in the same direction as the TPP's errors (e.g., both miss the same class of IO users), and show that the exponential tilting adjustment degrades rather than improves detection performance compared to the TPP alone.

Watch

Extended reading notes

Core claim

The central mechanism is the fusion of a temporal-behavioral anomaly score from a TPP with a language-aware evidence score from an LLM via exponential tilting. The key finding is that an LLM that performs poorly at zero-shot IO detection on its own still provides useful corrective signal when used as an evidence function to adjust a TPP that was trained on contaminated data. The LLM's judgments, even if noisy, carry information about IO characteristics that is complementary to what the TPP captures from temporal patterns alone, and the exponential tilting formula provides a principled way to combine them.

Load-bearing premise

The LLM-based evidence function provides signal that is genuinely complementary to the TPP score rather than correlated with the TPP's errors. The paper shows the LLM alone performs poorly at zero-shot IO detection, yet assumes its output is sufficient to correct the TPP through exponential tilting. If the LLM's judgments are systematically wrong in the same way the TPP is wrong, the adjustment could reinforce errors rather than correct them.

Editorial extensions

If this is right

  • Platforms could deploy TENSOR-style detectors that adapt to new IO campaigns without needing labeled examples of those specific campaigns, addressing the generalization bottleneck of supervised methods.
  • The evidence-function approach could generalize beyond IO detection to any anomaly detection task where a contaminated training set is unavoidable and a separate, imperfect signal source is available for post-hoc correction.
  • The finding that a weak LLM signal can correct a contaminated statistical model suggests a broader design pattern: using LLMs as evidence providers rather than primary classifiers in domains where they lack sufficient zero-shot accuracy.
  • Unsupervised optimization of the temperature parameter beta, which the paper identifies as promising but leaves open, could make the method fully label-free for deployment.
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, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 7 minor

Summary. This paper proposes TENSOR, an unsupervised anomaly detection method for Information Operations (IO) user detection on social media. The approach fuses two signals: (1) a Temporal Point Process (TPP) trained on user posting timestamps to capture behavioral anomaly, and (2) an LLM-derived evidence function that scores whether a user's post timeline resembles an IO account. These are combined via the EPHAD framework (Patra et al. [24]), which uses exponential tilting to post-hoc adjust the TPP anomaly score using the LLM evidence, motivated by the claim that this mitigates training data contamination (IO users present in the control training set). TENSOR is evaluated on five real-world IO datasets against three baselines (Clustering, BLOC, LLM-based zero-shot), showing improvements in AUPRC and other metrics. Ablation studies isolate the contributions of the TPP and LLM components, and sensitivity analyses examine different LLMs and temperature values.

Significance. The paper addresses a practically important problem (unsupervised IO user detection) and the multimodal fusion of temporal behavioral signals (TPP) with language signals (LLM evidence) via EPHAD is a reasonable and novel application of the post-hoc adjustment framework. The inclusion of ablation studies (Table 7), alternative fusion comparisons (Table 8), multi-LLM sensitivity (Table 9), and temperature sensitivity (Figure 4, Table 10) strengthens the empirical contribution. The code release is a positive sign of reproducibility. The five real-world datasets span different countries and contamination levels, providing a reasonable breadth of evaluation.

major comments (3)
  1. §1, §4.3, §6: The paper's central motivational claim is that EPHAD fusion of LLM evidence with TPP scores 'mitigates training data contamination.' However, the paper's own data does not substantiate this specific mechanism. From Table 1 and Table 7: Egypt has 47.5% IO users (219/461) and shows 11.2% relative AUPRC improvement from LLM fusion (0.6916→0.7690); China_1 has 1.9% IO users (537/28982) yet shows 19.3% improvement (0.4740→0.5654); UAE has 33.5% IO users (3337/9972) but only 6.9% improvement (0.8009→0.8562). If the mechanism were contamination mitigation, improvement should scale with contamination level, but no such correlation exists (if anything, the trend is inverted). This suggests the LLM provides complementary language signal unrelated to contamination correction. The performance results likely hold regardless of mechanism, but the 'contamination mitigation' framing — the纸
  2. §5, Evaluation Metrics: The threshold epsilon is tuned by maximizing F1 on a labeled validation set. While the paper states labels are used 'solely for performance evaluations,' threshold tuning on labeled data is a form of supervised model selection that weakens the 'unsupervised' framing for the precision/recall/F1 metrics. The AUPRC and AUC metrics (which are threshold-free) are more appropriate for evaluating the unsupervised claim, and the paper does report these. However, the F1 results in Table 4 and precision/recall in Tables 2-3 should be clearly caveated as requiring a labeled validation set, or the unsupervised threshold selection procedure should be elaborated.
  3. §4.3, Eq. (9): The evidence function T(t_i, c_i) is defined as the softmax of semantic similarities between the LLM output and reference texts ('IO account' vs. 'Control account'). The paper's own ablation (Table 7, 'TENSOR without TPP') shows the LLM alone performs poorly at zero-shot IO detection (average AUPRC 0.2857). This raises a concern: if the LLM's zero-shot judgments are weak, is the evidence function providing signal that is genuinely independent of the TPP score, or is it primarily adding noise that the exponential tilting happens to regularize? The paper should discuss whether the LLM evidence and TPP score are sufficiently independent for the EPHAD adjustment to be theoretically justified, or provide empirical evidence of their decorrelation.
minor comments (7)
  1. §5.1: The text states 'One outlier is Egypt, where Clustering performs better on AUPRC' but Table 6 shows TENSOR (0.7690) vs. Clustering (0.8232) on Egypt — the statement is correct but could note that this is the only dataset where Clustering wins on AUPRC.
  2. §4.3, Eq. (9): The similarity function sim(x, y) is mentioned but not defined until later (bge-m3-v2 reranker). A forward reference or brief definition at first use would improve readability.
  3. Table 10 caption: States 'temperature β = 0.5 for TENSOR results in the second column' but the first column is 'TENSOR with best β' — the caption should clarify how 'best β' was selected (presumably per-dataset optimization on validation AUPRC).
  4. §5.4: The text mentions 'it is consistently around 0.3 for gpt-oss-120B' but Figure 4 shows substantial variation across datasets (Egypt peaks near 0.1, UAE near 0.5). The claim of consistency should be softened.
  5. §3, Eq. (1): The threshold epsilon is introduced here but its determination is deferred to Section 5. A brief forward reference at this point would help the reader.
  6. Table 1: The total row sums (7,466 IO users, 59,574 control users) are useful but the percentage of IO users per dataset is not shown and would aid interpretation of contamination levels.
  7. §4.3: The computational complexity claim of O(M_i) for the LLM component is noted as 'O(1)' in footnote 5, but this is only true if the LLM processes the entire timeline in a single forward pass. This should be clarified, as LLM context length limits may require chunking.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful and constructive review. The three major comments are well-taken and have prompted us to reconsider aspects of our framing and presentation. We agree with two of the three comments in full and will revise accordingly; on the third (independence of LLM evidence and TPP scores), we agree the concern is legitimate and will add discussion plus empirical analysis, though we believe the existing ablation results already partially address it.

read point-by-point responses

  1. Referee: The paper's central motivational claim is that EPHAD fusion mitigates training data contamination, but the data does not substantiate this mechanism: improvement from LLM fusion does not correlate with contamination level (Egypt 47.5% IO → 11.2% improvement; China_1 1.9% IO → 19.3% improvement; UAE 33.5% IO → 6.9% improvement). This suggests the LLM provides complementary language signal unrelated to contamination correction.

    Authors: The referee is correct that our data does not substantiate a contamination-mitigation mechanism in the specific sense of improvement scaling with contamination level. We appreciate this careful analysis. Upon reflection, we agree that the contamination-mitigation framing is not supported by our experimental evidence and that a more accurate characterization is that the LLM evidence provides a complementary language-based signal that improves detection regardless of contamination level. We will revise the manuscript to reframe the motivation: rather than claiming EPHAD mitigates contamination specifically, we will describe the LLM evidence function as providing complementary signal from the content dimension that the TPP (trained on timestamps alone) cannot capture. The EPHAD framework remains theoretically motivated as a principled post-hoc adjustment that combines two imperfect but complementary signals. We will also add an explicit discussion of the contamination-improvement relationship the referee identifies, acknowledging that the absence of correlation rules out the contamination-mitigation mechanism and supporting the complementary-signal interpretation. revision: yes

  2. Referee: The threshold epsilon is tuned by maximizing F1 on a labeled validation set, which is a form of supervised model selection that weakens the 'unsupervised' framing for precision/recall/F1 metrics. The F1 results in Table 4 and precision/recall in Tables 2-3 should be caveated, or the unsupervised threshold selection procedure should be elaborated.

    Authors: The referee is correct. Tuning epsilon on a labeled validation set is a form of supervised model selection, and our current presentation does not adequately caveat this. We will make two changes. First, we will add an explicit caveat in Section 5 stating that while TENSOR's scoring is fully unsupervised (the TPP is trained without labels and the LLM evidence function requires no labels), the threshold epsilon for converting scores to binary labels is selected using a labeled validation set, making the precision, recall, and F1 metrics semi-supervised in the threshold-selection step. Second, we will emphasize that AUPRC and AUC, which are threshold-free and fully unsupervised, are the primary metrics for evaluating the unsupervised claim, and will note this explicitly when presenting Tables 2-4. We will also briefly mention that unsupervised threshold selection (e.g., modeling score distributions as two-component mixtures) is possible future work, as we already note for beta optimization. revision: yes

  3. Referee: The LLM's zero-shot judgments are weak (average AUPRC 0.2857 without TPP). Is the evidence function providing signal genuinely independent of the TPP score, or is it primarily adding noise that exponential tilting happens to regularize? The paper should discuss independence/decorrelation or provide empirical evidence.

    Authors: This is a legitimate and important concern. We agree that the EPHAD framework's theoretical justification assumes the evidence function provides signal not already captured by the base detector. We will address this in two ways. First, we will add discussion noting that the TPP score is computed solely from timestamps (behavioral patterns), while the LLM evidence function is computed from post content (language patterns), so the two signals are derived from fundamentally different data modalities, which provides a structural basis for independence. Second, we will add an empirical analysis computing the correlation between the TPP scores and LLM evidence scores across users on each dataset. We expect this to show low correlation, supporting the claim that the LLM evidence provides genuinely complementary signal. We also note that the existing ablation in Table 7 already provides relevant evidence: 'TENSOR with random evidence' (uniform noise replacing the LLM) performs worse than 'TENSOR without LLM' on average (0.6299 vs. 0.6506), while TENSOR with the actual LLM evidence performs substantially better (0.7154). If the LLM evidence were merely noise that exponential tilting regularizes, the random-evidence ablation should perform comparably to the real-evidence version. The fact that real LLM evidence significantly outperforms random noise demonstrates that the LLM provides useful signal beyond what the TPP captures, even though the LLM alone is a weak detector. We will make this argument explicit in the revised manuscript. revision: partial

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: TENSOR's core result is independently grounded in external datasets and benchmarks, with one minor self-citation that is not load-bearing.

full rationale

The paper's central claim — that fusing TPP-based behavioral anomaly scores with LLM-derived evidence scores via EPHAD (Eq. 11) outperforms unsupervised baselines on five real-world IO datasets — is evaluated against external benchmarks (Tables 2–6, 7–10) with no fitted parameter renamed as a prediction. The EPHAD framework (Eq. 5–7) is cited from Patra et al. [24], an external NeurIPS 2025 paper with no author overlap, and is applied as a pre-existing mathematical tool (exponential tilting solving a KL-regularized optimization, Eq. 6). The TPP model SAHP is cited from Zhang et al. [40], also external. The evidence function T(t_i, c_i) (Eq. 9–10) is a novel construction converting LLM outputs to probability scores via semantic similarity mapping; it is not defined in terms of the TPP output it adjusts, so there is no self-defitional circularity. The temperature β=0.5 is inherited from [24] (not fit to the target data), and the threshold ε is tuned on a validation set (standard practice, not circular). The ablation in Table 7 ('TENSOR without TPP') showing the LLM alone performs poorly (AUPRC 0.2857) is evidence against, not for, circularity — it demonstrates the two components provide complementary signal. The reader's concern about whether 'contamination mitigation' is the correct causal mechanism for the improvement is a correctness/validity issue, not a circularity issue: the performance results themselves (TENSOR > baselines) are measured on held-out test data and are not forced by construction. No step in the derivation chain reduces to its own inputs by definition or by self-citation. The one self-citation present (SAHP [40], Zhang et al.) is for a standard TPP model choice and is not load-bearing for the central fusion claim. Score: 1.

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

The paper does not invent new physical or mathematical entities. It introduces a new function (the evidence function) and a new system (TENSOR), but these are composed of existing components (TPP, LLM, EPHAD). The free parameters (beta, epsilon, N) are standard for this type of model, though the tuning of epsilon on labeled validation data is a notable caveat for an unsupervised method.

free parameters (3)
  • beta (temperature) = 0.5
    Inherited from EPHAD paper [24]. The paper acknowledges this is suboptimal and that a lower beta (~0.3) improves performance, but leaves unsupervised optimization of beta as future work.
  • epsilon (threshold) = maximized F1 on validation set
    Used to decide the IO/control label boundary. Adjusted by maximizing F1 on the validation set, which requires labels and weakens the unsupervised claim.
  • N (number of LLM samples) = 5
    Number of samples drawn for the evidence function to compute the mean. Chosen without stated justification.
assumptions (3)
  • domain assumption IO users are a very small fraction of all online users and have unique temporal behavioral and language patterns.
    Stated in the abstract and introduction. This is the foundational assumption for framing the problem as anomaly detection.
  • standard math Anomaly detection models trained on contaminated data can be corrected at test time using a predefined evidence function (EPHAD).
    Invoked in Section 4.2, citing Patra et al. [24] and Korbak et al. [13]. The paper relies on this framework for its core fusion mechanism.
  • domain assumption LLM responses, when converted to quantitative scores via semantic similarity, provide useful evidence for distinguishing IO users from control users.
    This underpins the design of the evidence function in Equation 9. The paper's own ablation shows the LLM alone is weak, but assumes it is still useful as evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Unsupervised Anomaly Detection of Information Operations Users via Behavioral and Language Patterns." pith.science (2026). https://pith.science/paper/WTYNF63W

@misc{pith2026260705855,
  author       = {Pith},
  title        = {Pith review of: Unsupervised Anomaly Detection of Information Operations Users via Behavioral and Language Patterns},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WTYNF63W}},
  note         = {Machine review of arXiv:2607.05855}
}
read the original abstract

Information Operations on social media networks have been identified as a significant threat to democracy and modern society, but they are challenging and expensive to detect by humans. Existing supervised IO detection methods fail to capture the dynamic nature of evolving IO user behavior, while existing unsupervised approaches rely on oversimplified assumptions of coordination among IO users that may not exist in practice. To overcome the limitations of existing methods, we formulate IO user detection as an anomaly detection problem and propose a novel unsupervised IO user detection approach called Temporal-bEhavior-laNguage Signals for information Operation Recognition (TENSOR), which leverages multimodal data, including temporal online user behavior, such as message posting activities, and the textual content of the messages. The motivation is that IO users are typically a very small fraction of all online users and have unique temporal behavioral and language patterns. Specifically, we train a Temporal Point Process (TPP) to capture abnormal temporal behavioral patterns of IO users because they are known to behave in a coordinated manner for IO campaigns. We further introduce a novel evidence function that converts LLM responses, which are generated from user post timelines, into quantitative scores to adjust the TPP outputs for better IO user detection. Experimental results show that TENSOR outperforms the baselines on five real-world IO datasets. Code is available at https://github.com/xiuzhenzhang/TENSOR.

Figures

Figures reproduced from arXiv: 2607.05855 by the authors.

Figure 1
Figure 1. Architecture of Temporal-bEhavior-laNguage Signals for information Operation Recognition (TENSOR). 4.3 Temporal-bEhavior-laNguage Signals for information Operation Recognition In this section, we propose Temporal-bEhavior-laNguage Signals for information Operation Recognition (TENSOR), which is sketched in [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Prompts for zero-shot IO user detection in [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Impact of data imbalance on Clustering’s performance. Data points represent control users (C0) and IO users (C1). Results indicate that significant data imbalance shifts cluster centroids with some random seeds, leading to performance inconsistency. users. This instability results in inconsistent performance, particularly on highly imbalanced datasets such as China_1 and Russia_1. 5.2 The impact of behavioral and la… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: AUPRC of TENSOR on all five IO datasets with different temperatures from 0.05 to 2 [PITH_FULL_IMAGE:figures/full_fig_p015_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

  1. [24]

    In: NeurIPS 2025 (2025)

    Patra, S., Taieb, S.B.: An Evidence-Based Post-Hoc Adjustment Framework for Anomaly Detection Under Data Contamination. In: NeurIPS 2025 (2025)

    work page 2025

  2. [1]

    Proceedings of the International AAAI Conference on Web and Social Media13, 15–25 (2019)

    Addawood, A., Badawy, A., Lerman, K., Ferrara, E.: Linguistic Cues to Deception: Identifying Political Trolls on Social Media. Proceedings of the International AAAI Conference on Web and Social Media13, 15–25 (2019)

    work page 2019

  3. [2]

    Science Advances6(30) (2020)

    Alizadeh, M., Shapiro, J.N., Buntain, C., Tucker, J.A.: Content-based fea- tures predict social media influence operations. Science Advances6(30) (2020)

    work page 2020

  4. [3]

    In: ACL 2024

    Chen, J., Xiao, S., Zhang, P., Luo, K., Lian, D., Liu, Z.: M3-Embedding: Multi-Linguality, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation. In: ACL 2024. pp. 2318–2335 (2024)

    work page 2024

  5. [4]

    (eds.): An Introduction to the Theory of Point Processes Volume I: Elementary Theory and Methods

    Daley, D.J., Vere-Jones, D. (eds.): An Introduction to the Theory of Point Processes Volume I: Elementary Theory and Methods. Probability and its Applications, Springer, 2 edn. (2003)

    work page 2003

  6. [5]

    In: ICML 2006

    Davis, J., Goadrich, M.: The relationship between Precision-Recall and ROC curves. In: ICML 2006. pp. 233–240 (2006)

    work page 2006

  7. [6]

    EPJ Data Science 12(1), 46 (2023)

    Ezzeddine, F., Ayoub, O., Giordano, S., Nogara, G., Sbeity, I., Ferrara, E., Luceri, L.: Exposing influence campaigns in the age of LLMs: a behavioral- based AI approach to detecting state-sponsored trolls. EPJ Data Science 12(1), 46 (2023)

    work page 2023

  8. [7]

    World Economic Forum, 19th ed edn

    Forum, W.E.: The global risks report 2024: insight report. World Economic Forum, 19th ed edn. (2024), oCLC: 1499197716

    work page 2024

Show all 43 references
  1. [8]

    In: WWW 2023 Companion

    Haider, S., Luceri, L., Deb, A., Badawy, A., Peng, N., Ferrara, E.: Detecting Social Media Manipulation in Low-Resource Languages. In: WWW 2023 Companion. pp. 1358–1364 (2023)

    2023

  2. [9]

    In: WebSci 2020

    Im, J., Chandrasekharan, E., Sargent, J., Lighthammer, P., Denby, T., Bhargava, A., Hemphill, L., Jurgens, D., Gilbert, E.: Still out there: Modeling and Identifying Russian Troll Accounts on Twitter. In: WebSci 2020. pp. 1–10 (2020)

    2020

  3. [10]

    Jachim, P., Sharevski, F., Treebridge, P.: TrollHunter [Evader]: Automated Detection [Evasion] of Twitter Trolls During the COVID-19 Pandemic (2020), arXiv:2012.02586

    2020 arXiv

  4. [11]

    Transactions on Machine Learning Research (2025)

    Ke, Z., Jiao, F., Ming, Y., Nguyen, X.P., Xu, A., Long, D.X., Li, M., Qin, C., Wang, P., Savarese, S., Xiong, C., Joty, S.: A Survey of Frontiers in LLM Reasoning: Inference Scaling, Learning to Reason, and Agentic Systems. Transactions on Machine Learning Research (2025)

    2025

  5. [12]

    In: Proceedings of the ACM Web Conference 2023

    Kong,Q.,Calderon,P.,Ram,R.,Boichak,O.,Rizoiu,M.A.:Interval-censored Transformer Hawkes: Detecting Information Operations using the Reaction of Social Systems. In: Proceedings of the ACM Web Conference 2023. pp. 1813–1821. WWW ’23 (2023)

    2023

  6. [13]

    In: EMNLP 2022

    Korbak, T., Perez, E., Buckley, C.: RL with KL penalties is better viewed as Bayesian inference. In: EMNLP 2022. pp. 1083–1091. Association for Computational Linguistics (2022) IO User Detection via Behavior and Language Patterns 17

    2022

  7. [14]

    Kumar, R., Goddu, B., Saha, S., Jatowt, A.: Silver Lining in the Fake News Cloud: Can Large Language Models Help Detect Misinformation? IEEE Transactions on Artificial Intelligence6(1), 14–24 (2025)

    2025

  8. [15]

    TMLR (2022)

    Lin, H., Wu, L., Zhao, G., Pai, L., Li, S.Z.: Exploring Generative Neural Temporal Point Process. TMLR (2022)

    2022

  9. [16]

    In: Proceedings of the 38th International Conference on Machine Learning

    Liu, S., Hauskrecht, M.: Event Outlier Detection in Continuous Time. In: Proceedings of the 38th International Conference on Machine Learning. pp. 6793–6803. PMLR (2021)

    2021

  10. [17]

    In: Companion Proceedings of the ACM Web Conference 2024

    Luceri, L., Boniardi, E., Ferrara, E.: Leveraging Large Language Models to Detect Influence Campaigns on Social Media. In: Companion Proceedings of the ACM Web Conference 2024. pp. 1459–1467 (2024)

    2024

  11. [18]

    Proceedings of the International AAAI Conference on Web and Social Media14, 417–427 (2020)

    Luceri, L., Giordano, S., Ferrara, E.: Detecting Troll Behavior via Inverse Reinforcement Learning: A Case Study of Russian Trolls in the 2016 US Election. Proceedings of the International AAAI Conference on Web and Social Media14, 417–427 (2020)

    2016

  12. [19]

    IEEE Access12, 15902–15916 (2024)

    Malik, U., Bernard, S., Pauchet, A., Chatelain, C., Picot-Clémente, R., Cortinovis, J.: Pseudo-Labeling With Large Language Models for Multi- Label Emotion Classification of French Tweets. IEEE Access12, 15902–15916 (2024)

    2024

  13. [20]

    Proceedings of the AAAI Conference on Artificial Intelligence39(27), 28258–28266 (2025)

    Minici, M., Luceri, L., Fabbri, F., Ferrara, E.: IOHunter: Graph Foundation Model to Uncover Online Information Operations. Proceedings of the AAAI Conference on Artificial Intelligence39(27), 28258–28266 (2025)

    2025

  14. [21]

    EPJ Data Science12(1), 33 (2023)

    Nwala, A.C., Flammini, A., Menczer, F.: A language framework for modeling social media account behavior. EPJ Data Science12(1), 33 (2023)

    2023

  15. [22]

    OpenAI: OpenAI o1 System Card (2024), arXiv:2412.16720

    2024 arXiv

  16. [23]

    OpenAI: gpt-oss-120b & gpt-oss-20b Model Card (2025), arXiv:2508.10925

    2025 arXiv

  17. [25]

    In: Proceedings of the 40th International Conference on Machine Learning

    Perini, L., Bürkner, P.C., Klami, A.: Estimating the Contamination Factor’s Distribution in Unsupervised Anomaly Detection. In: Proceedings of the 40th International Conference on Machine Learning. pp. 27668–27679. PMLR (2023)

    2023

  18. [26]

    In: CVPR

    Qi, P., Yan, Z., Hsu, W., Lee, M.L.: SNIFFER: Multimodal Large Language Model for Explainable Out-of-Context Misinformation Detection. In: CVPR. pp. 13052–13062 (2024)

    2024

  19. [27]

    In: Proceedings of the 39th International Conference on Machine Learning

    Qiu, C., Li, A., Kloft, M., Rudolph, M., Mandt, S.: Latent Outlier Exposure for Anomaly Detection with Contaminated Data. In: Proceedings of the 39th International Conference on Machine Learning. pp. 18153–18167. PMLR (2022)

    2022

  20. [28]

    Proceedings of the International AAAI Conference on Web and Social Media 19, 2567–2574 (2025)

    Seckin, O.C., Pote, M., Nwala, A.C., Yin, L., Luceri, L., Flammini, A., Menczer, F.: Labeled Datasets for Research on Information Operations. Proceedings of the International AAAI Conference on Web and Social Media 19, 2567–2574 (2025)

    2025

  21. [29]

    In: NeurIPS 2021 (2021) 18 S

    Shchur, O., Turkmen, A.C., Januschowski, T., Gasthaus, J., Günnemann, S.: Detecting Anomalous Event Sequences with Temporal Point Processes. In: NeurIPS 2021 (2021) 18 S. Liu et al

    2021

  22. [30]

    In: IJCAI 2021

    Shchur, O., Türkmen, A.C., Januschowski, T., Günnemann, S.: Neural Temporal Point Processes: A Review. In: IJCAI 2021. pp. 4585–4593 (2021)

    2021

  23. [31]

    Pro- ceedings of the National Academy of Sciences118(4) (2021)

    Smith, S.T., Kao, E.K., Mackin, E.D., Shah, D.C., Simek, O., Rubin, D.B.: Automatic detection of influential actors in disinformation networks. Pro- ceedings of the National Academy of Sciences118(4) (2021)

    2021

  24. [32]

    Team, G..: GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models (2025), arXiv:2508.06471

    2025 arXiv

  25. [33]

    Team, Q.: Qwen3 technical report (2025),https://arxiv.org/abs/2505. 09388

    2025

  26. [34]

    In: HCI International 2023 - Late Breaking Papers

    Tekumalla, R., Banda, J.M.: Leveraging Large Language Models and Weak Supervision for Social Media Data Annotation: An Evaluation Using COVID- 19 Self-reported Vaccination Tweets. In: HCI International 2023 - Late Breaking Papers. pp. 356–366. Springer Nature Switzerland (2023)

    2023

  27. [35]

    Social Science Computer Review43(6), 1181–1195 (2025)

    Törnberg, P.: Large Language Models Outperform Expert Coders and Su- pervised Classifiers at Annotating Political Social Media Messages. Social Science Computer Review43(6), 1181–1195 (2025)

    2025

  28. [36]

    In: CCSW 2020

    Vargas, L., Emami, P., Traynor, P.: On the Detection of Disinformation Campaign Activity with Network Analysis. In: CCSW 2020. pp. 133–146 (2020)

    2020

  29. [37]

    Proceedings of the International AAAI Conference on Web and Social Media18, 1557–1570 (2024)

    Vishnuprasad, P.S., Nogara, G., Cardoso, F., Cresci, S., Giordano, S., Luceri, L.: Tracking Fringe and Coordinated Activity on Twitter Leading Up to the US Capitol Attack. Proceedings of the International AAAI Conference on Web and Social Media18, 1557–1570 (2024)

    2024

  30. [38]

    In: Advances in Neural Information Processing Systems

    Wang, S., Zeng, Y., Liu, X., Zhu, E., Yin, J., Xu, C., Kloft, M.: Effective End- to-end Unsupervised Outlier Detection via Inlier Priority of Discriminative Network. In: Advances in Neural Information Processing Systems. vol. 32 (2019)

    2019

  31. [39]

    Transactions on Machine Learning Research (2022)

    Yoon, J., Sohn, K., Li, C.L., Arik, S.O., Lee, C.Y., Pfister, T.: Self-supervise, Refine, Repeat: Improving Unsupervised Anomaly Detection. Transactions on Machine Learning Research (2022)

    2022

  32. [40]

    In: Proceedings of the 37th International Conference on Machine Learning

    Zhang, Q., Lipani, A., Kirnap, O., Yilmaz, E.: Self-Attentive Hawkes Process. In: Proceedings of the 37th International Conference on Machine Learning. pp. 11183–11193. PMLR (2020)

    2020

  33. [41]

    In: ICDM

    Zhang, S., Zhou, C., Zhang, P., Liu, Y., Li, Z., Chen, H.: Multiple Hypothesis Testing for Anomaly Detection in Multi-type Event Sequences. In: ICDM. pp. 808–817 (2023)

    2023

  34. [42]

    In: WWW 2024 Companion

    Zhang, Y., Sharma, K., Du, L., Liu, Y.: Toward Mitigating Misinformation and Social Media Manipulation in LLM Era. In: WWW 2024 Companion. pp. 1302–1305. Association for Computing Machinery (2024)

    2024

  35. [43]

    Zhou, X., Sharma, A., Zhang, A.X., Althoff, T.: Correcting misinformation on social media with a large language model (2026), arXiv:2403.11169

    2026

Pith tools

Reviewed July 8, 2026 · model on record in the stance chip above.