pith. sign in

arxiv: 2604.21767 · v1 · submitted 2026-04-23 · 💻 cs.CL · cs.SI

Misinformation Span Detection in Videos via Audio Transcripts

Breno Matos , Rennan C. Lima , Savvas Zannettou , Fabricio Benevenuto , Rodrygo L.T. Santos This is my paper

Pith reviewed 2026-05-09 22:03 UTC · model grok-4.3

classification 💻 cs.CL cs.SI
keywords misinformation detectionvideo analysisaudio transcriptsspan detectionnatural language processinglanguage modelsfact checking
0
0 comments X

The pith

Audio transcripts let language models locate misinformation to specific video segments at F1 0.68.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper creates two new datasets from over 500 videos containing more than 2,400 annotated segments where fact-checked misinformation claims appear. It transcribes each video's audio into text and trains classifiers on those transcripts to identify the exact spans responsible for the misleading content. Prior work only labeled entire videos as true or false, but this span-level approach supplies the missing detail on which claims drive the misinformation. The method reaches an F1 score of 0.68 using current language models and releases the datasets, transcripts, audio, and videos for further use.

Core claim

By converting video audio to text and labeling the segments that contain verified misinformation claims, two datasets are built that support training models to detect the precise portions of a video responsible for its misleading nature. Classifiers built with state-of-the-art language models applied to these transcripts identify the misinformation spans at an F1 score of 0.68.

What carries the argument

Misinformation span detection on audio transcripts using language-model classifiers trained on annotated video segments.

Load-bearing premise

Human annotations correctly and consistently mark the exact spans that carry misinformation, and transcripts alone contain enough information to detect those claims without visuals or other context.

What would settle it

New independent annotators marking the same videos produce substantially different span boundaries, or models trained on transcripts miss claims that only become clear when the video image is viewed.

Figures

Figures reproduced from arXiv: 2604.21767 by Breno Matos, Fabricio Benevenuto, Rennan C. Lima, Rodrygo L.T. Santos, Savvas Zannettou.

Figure 1
Figure 1. Figure 1: Example of a fact-checked video with pointers to [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of our methodology regarding the BOL4Y dataset [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Monthly sum of misinformation claims. Vertical lines signal important events during Bolsonaro’s administration. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Temporal analysis of the performance of our classifiers. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
read the original abstract

Online misinformation is one of the most challenging issues lately, yielding severe consequences, including political polarization, attacks on democracy, and public health risks. Misinformation manifests in any platform with a large user base, including online social networks and messaging apps. It permeates all media and content forms, including images, text, audio, and video. Distinctly, video-based misinformation represents a multifaceted challenge for fact-checkers, given the ease with which individuals can record and upload videos on various video-sharing platforms. Previous research efforts investigated detecting video-based misinformation, focusing on whether a video shares misinformation or not on a video level. While this approach is useful, it only provides a limited and non-easily interpretable view of the problem given that it does not provide an additional context of when misinformation occurs within videos and what content (i.e., claims) are responsible for the video's misinformation nature. In this work, we attempt to bridge this research gap by creating two novel datasets that allow us to explore misinformation detection on videos via audio transcripts, focusing on identifying the span of videos that are responsible for the video's misinformation claim (misinformation span detection). We present two new datasets for this task. We transcribe each video's audio to text, identifying the video segment in which the misinformation claims appears, resulting in two datasets of more than 500 videos with over 2,400 segments containing annotated fact-checked claims. Then, we employ classifiers built with state-of-the-art language models, and our results show that we can identify in which part of a video there is misinformation with an F1 score of 0.68. We make publicly available our annotated datasets. We also release all transcripts, audio and videos.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces the task of misinformation span detection in videos using audio transcripts. It creates two novel datasets from over 500 videos with more than 2,400 annotated segments containing fact-checked claims, transcribes the audio to text, and trains classifiers with state-of-the-art language models to identify the specific spans responsible for misinformation, reporting an F1 score of 0.68. The datasets, transcripts, audio, and videos are released publicly.

Significance. If the central results hold after addressing validation gaps, the work could meaningfully advance fine-grained, interpretable misinformation detection beyond coarse video-level classification by localizing claims within transcripts. The public release of the annotated datasets and associated media is a clear strength that would support reproducibility and follow-on research in computational linguistics and fact-checking.

major comments (3)
  1. [Abstract] Abstract: The abstract states an F1 score of 0.68 but supplies no information on dataset construction details, annotation process, model choices, baselines, or evaluation splits, preventing verification that the numbers support the claim.
  2. [Dataset sections] Dataset sections: The description of annotating fact-checked claims into video segments reports neither inter-annotator agreement nor any details on annotation guidelines, number of annotators, or consistency checks. This is load-bearing because the F1 score interpretation requires the >2,400 segments to constitute reliable ground truth.
  3. [Experiments/Results] Experiments/Results: No ablation or analysis is provided to show that performance derives from transcript features alone rather than visual/contextual cues, leaving the claim that 'audio transcripts alone' suffice untested.
minor comments (2)
  1. [Abstract] Abstract: The sentence 'identifying the span of videos that are responsible for the video's misinformation claim' is repetitive and could be rephrased for clarity.
  2. [Abstract] Abstract: Consider specifying the exact number of videos and segments early in the abstract to give readers immediate scale.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback on our manuscript. We address each of the major comments point by point below, indicating the revisions we plan to make to strengthen the paper.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The abstract states an F1 score of 0.68 but supplies no information on dataset construction details, annotation process, model choices, baselines, or evaluation splits, preventing verification that the numbers support the claim.

    Authors: We agree that the abstract could benefit from additional context to support the reported results. In the revised version, we will expand the abstract to briefly mention the dataset construction (two novel datasets from over 500 videos with more than 2,400 annotated segments), the annotation of fact-checked claims, the use of state-of-the-art language models for classification, and the evaluation leading to an F1 score of 0.68. Specific details on baselines and splits will be cross-referenced to the main text. revision: yes

  2. Referee: [Dataset sections] Dataset sections: The description of annotating fact-checked claims into video segments reports neither inter-annotator agreement nor any details on annotation guidelines, number of annotators, or consistency checks. This is load-bearing because the F1 score interpretation requires the >2,400 segments to constitute reliable ground truth.

    Authors: We recognize the critical need for detailed annotation information to ensure the reliability of the ground truth. We will revise the dataset sections to provide more comprehensive details on the annotation process, including the number of annotators, the guidelines followed for mapping fact-checked claims to video segments, and any procedures used to ensure consistency. Inter-annotator agreement was not calculated during the original annotation; we will acknowledge this limitation in the revised manuscript and discuss its implications for the results. revision: partial

  3. Referee: [Experiments/Results] Experiments/Results: No ablation or analysis is provided to show that performance derives from transcript features alone rather than visual/contextual cues, leaving the claim that 'audio transcripts alone' suffice untested.

    Authors: The proposed method and classifiers operate exclusively on the audio transcripts using language models, without incorporating any visual or additional contextual features from the videos. As such, the F1 score of 0.68 is derived solely from transcript-based features. We will update the experiments and results section to explicitly clarify this aspect and include a brief discussion or note to confirm that no visual cues are utilized in the classification process. revision: yes

Circularity Check

0 steps flagged

No circularity in derivation chain

full rationale

The paper describes a conventional supervised learning pipeline: transcribe video audio, have humans annotate misinformation claim spans to create new datasets (>500 videos, >2400 segments), then train language model classifiers on the transcripts to predict spans, reporting an empirical F1 of 0.68. No equations, derivations, or first-principles results are claimed. No self-definitional steps, fitted inputs renamed as predictions, load-bearing self-citations, imported uniqueness theorems, or ansatzes smuggled via citation appear in the provided text. The central result is an out-of-sample performance metric on newly collected external data, making the chain self-contained without reduction to its own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the quality of human annotations for spans and the assumption that transcripts suffice for claim detection; no free parameters or invented entities are mentioned.

axioms (1)
  • domain assumption Audio transcripts of videos accurately capture the spoken claims needed for misinformation identification
    The datasets are built by transcribing audio and annotating segments based on those transcripts.

pith-pipeline@v0.9.0 · 5630 in / 1179 out tokens · 42001 ms · 2026-05-09T22:03:00.105884+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

79 extracted references · 79 canonical work pages · 1 internal anchor

  1. [1]

    Global environmental change , year=

    Priming critical thinking: Simple interventions limit the influence of fake news about climate change on Facebook , author=. Global environmental change , year=

    work page

  2. [2]

    Two faces: Editorial on BJP Govt’s manipulative ways to spread fake news , url=

    Board, The Editorial , year=. Two faces: Editorial on BJP Govt’s manipulative ways to spread fake news , url=. Google | Two faces: Editorial on BJP govt’s manipulative ways to spread fake news - Telegraph India , publisher=

    work page

  3. [3]

    The India Eye

    Jignesh Patel, AltNews.in , year=. Namo app promotes fake news factory “The India Eye” and users can’t block it even if they want to , url=. Scroll.in , publisher=

    work page

  4. [4]

    Is Narendra Modi’s namo app spreading fake news? A journalist finds a troubling pattern , url=

    Aria Thaker, qz.com , year=. Is Narendra Modi’s namo app spreading fake news? A journalist finds a troubling pattern , url=. Scroll.in , publisher=

    work page

  5. [5]

    Narendra Modi app has a fake news problem , url=

    Bansal, Samarth , year=. Narendra Modi app has a fake news problem , url=. HuffPost , publisher=

    work page

  6. [6]

    Brasil de Fato , author=

    Brazil’s Federal Public Ministry demands compensation from senator Damares Alves and the state for Marajó Archipelago for spreading fake news , url=. Brasil de Fato , author=. 2023 , month=

    work page 2023

  7. [7]

    Bolsonaro directly spread disinformation on Brazil’s voting system, police report says , url=

    Brito, Ricardo , year=. Bolsonaro directly spread disinformation on Brazil’s voting system, police report says , url=. Reuters , publisher=

    work page

  8. [8]

    Folha de S.Paulo , author=

    Instagram classifies Bolsonaro’s coronavirus post as fake news and hides it , url=. Folha de S.Paulo , author=. 2020 , month=

    work page 2020

  9. [9]

    Facebook shuts down network of fake accounts tied to employees of Brazil’s Bolsonaro and sons , url=

    Thornton, Chandler and Pedroso, Rodrigo , year=. Facebook shuts down network of fake accounts tied to employees of Brazil’s Bolsonaro and sons , url=. CNN , publisher=

    work page

  10. [10]

    Mensagem de Bolsonaro Com conteúdos falsos FOI Replicada por Aliados e Abasteceu Canais Extremistas , url=

    Caetano, Guilherme and Couto, Marlen , year=. Mensagem de Bolsonaro Com conteúdos falsos FOI Replicada por Aliados e Abasteceu Canais Extremistas , url=. O Globo , publisher=

    work page

  11. [11]

    Police call for Bolsonaro to be charged for spreading Covid misinformation , url=

    Phillips, Tom , year=. Police call for Bolsonaro to be charged for spreading Covid misinformation , url=. The Guardian , publisher=

    work page

  12. [12]

    lit that fire

    Jalonick, Mary Clare and Tucker, Eric and Balsamo, Michael and Merchant, Nomaan and Colvin, Jill and Amiri, Farnoush , year=. Jan. 6 report: Trump “lit that fire” of capitol insurrection , url=. AP News , publisher=

    work page

  13. [13]

    unhinged

    Zengerle, Patricia and Cowan, Richard and Chiacu, Doina , year=. Trump incited Jan. 6 attack after “unhinged” White House meeting, panel told , url=. Reuters , publisher=

    work page

  14. [14]

    Harvard Kennedy School Misinformation Review , year=

    Twitter flagged Donald Trump’s tweets with election misinformation: They continued to spread both on and off the platform , author=. Harvard Kennedy School Misinformation Review , year=

    work page

  15. [15]

    Macdonald, Maggie and Brown, Megan A. , year=. Analysis | Republicans are increasingly sharing misinformation, research finds , url=. The Washington Post , publisher=

    work page

  16. [16]

    arXiv:2008.09144 , year=

    Ptt5: Pretraining and validating the t5 model on brazilian portuguese data , author=. arXiv:2008.09144 , year=

    work page arXiv 2008

  17. [17]

    Exploring the limits of transfer learning with a unified text-to-text transformer , author=

    work page

  18. [18]

    Wagner, Jorge and Wilkens, Rodrigo and Idiart, Marco and Villavicencio, Aline , year =

    work page

  19. [19]

    9th BRACIS, Brazil , year =

    F. 9th BRACIS, Brazil , year =

    work page

  20. [20]

    ICML , year=

    Robust speech recognition via large-scale weak supervision , author=. ICML , year=

    work page

  21. [21]

    Machine learning , year=

    An instance level analysis of data complexity , author=. Machine learning , year=

    work page

  22. [22]

    ICWSM , year=

    Market forces: Quantifying the role of top credible ad servers in the fake news ecosystem , author=. ICWSM , year=

    work page

  23. [23]

    Proceedings of the International AAAI Conference on Web and Social Media , year=

    HateMM: A Multi-Modal Dataset for Hate Video Classification , author=. Proceedings of the International AAAI Conference on Web and Social Media , year=

    work page

  24. [24]

    arXiv:2107.00941 , year=

    Misinformation detection on YouTube using video captions , author=. arXiv:2107.00941 , year=

    work page arXiv

  25. [25]

    ICWSM , author=

    YouNICon: YouTube’s CommuNIty of Conspiracy Videos , volume=. ICWSM , author=. doi:10.1609/icwsm.v17i1.22218 , abstractNote=

    work page doi:10.1609/icwsm.v17i1.22218

  26. [26]

    and Yan, Guanhua and Thai, My T

    Nguyen, Nam P. and Yan, Guanhua and Thai, My T. and Eidenbenz, Stephan , title =. Proceedings of the 4th Annual ACM Web Science Conference , series =. 2012 , publisher =

    work page 2012

  27. [27]

    Poynter , author=

    IFCN is heartened by A nobel peace prize nomination for the work of the global fact-checking community , url=. Poynter , author=. 2021 , month=

    work page 2021

  28. [28]

    A Multimodal Misinformation Detector for COVID-19 Short Videos on TikTok , year=

    Shang, Lanyu and Kou, Ziyi and Zhang, Yang and Wang, Dong , booktitle=. A Multimodal Misinformation Detector for COVID-19 Short Videos on TikTok , year=

    work page

  29. [29]

    Todas as declarações de Bolsonaro, checadas , year =

    work page

  30. [30]

    AAAI'23 , year=

    FakeSV: A multimodal benchmark with rich social context for fake news detection on short video platforms , author=. AAAI'23 , year=

    work page

  31. [31]

    Interfaces (GUI) , year=

    Recurrent convolutional strategies for face manipulation detection in videos , author=. Interfaces (GUI) , year=

    work page

  32. [32]

    WWW , year=

    Deepfake videos in the wild: Analysis and detection , author=. WWW , year=

    work page

  33. [33]

    2019 ICMI , year=

    Towards automatic detection of misinformation in online medical videos , author=. 2019 ICMI , year=

    work page 2019

  34. [34]

    2020 IEEE 7th DSAA , year=

    Multimodal multi-image fake news detection , author=. 2020 IEEE 7th DSAA , year=

    work page 2020

  35. [35]

    IEEE ICDM , year=

    Exploiting multi-domain visual information for fake news detection , author=. IEEE ICDM , year=

    work page

  36. [36]

    SpotFake: A Multi-modal Framework for Fake News Detection , year=

    Singhal, Shivangi and Shah, Rajiv Ratn and Chakraborty, Tanmoy and Kumaraguru, Ponnurangam and Satoh, Shin'ichi , booktitle=. SpotFake: A Multi-modal Framework for Fake News Detection , year=

    work page

  37. [37]

    , title =

    Huh, Minyoung and Liu, Andrew and Owens, Andrew and Efros, Alexei A. , title =. ECCV , month =

    work page

  38. [38]

    2022 , author =

    ANTi-Vax: a novel Twitter dataset for COVID-19 vaccine misinformation detection , journal =. 2022 , author =

    work page 2022

  39. [39]

    arXiv:2010.08768 , year=

    ArCOV19-rumors: Arabic COVID-19 twitter dataset for misinformation detection , author=. arXiv:2010.08768 , year=

    work page arXiv 2010

  40. [40]

    Neural Computing and Applications , year=

    Deep learning for fake news detection on Twitter regarding the 2019 Hong Kong protests , author=. Neural Computing and Applications , year=

    work page 2019

  41. [41]

    WWW 2021 , location =

    Pelrine, Kellin and Danovitch, Jacob and Rabbany, Reihaneh , title =. WWW 2021 , location =. 2021 , publisher =

    work page 2021

  42. [42]

    International Journal of Advanced Computer Science and Applications , year=

    Fake news detection in Arabic tweets during the COVID-19 pandemic , author=. International Journal of Advanced Computer Science and Applications , year=

    work page

  43. [43]

    WICO Text: A Labeled Dataset of Conspiracy Theory and 5G-Corona Misinformation Tweets , year =

    Pogorelov, Konstantin and Schroeder, Daniel Thilo and Filkukov\'. WICO Text: A Labeled Dataset of Conspiracy Theory and 5G-Corona Misinformation Tweets , year =. Proceedings of the 2021 Workshop on Open Challenges in Online Social Networks , series =

    work page 2021

  44. [44]

    2022 , author =

    A heuristic-driven uncertainty based ensemble framework for fake news detection in tweets and news articles , journal =. 2022 , author =

    work page 2022

  45. [45]

    IEEE Signal Processing Letters , year=

    Generalization of spectral flatness measure for non-gaussian linear processes , author=. IEEE Signal Processing Letters , year=

    work page

  46. [46]

    2023 , author =

    Hindi fake news detection using transformer ensembles , journal =. 2023 , author =

    work page 2023

  47. [47]

    Mathematics , YEAR =

    Truica, Ciprian-Octavian and Apostol, Elena-Simona , TITLE =. Mathematics , YEAR =

    work page

  48. [48]

    BART : Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension

    Lewis, Mike and Liu, Yinhan and Goyal, Naman and Ghazvininejad, Marjan and Mohamed, Abdelrahman and Levy, Omer and Stoyanov, Veselin and Zettlemoyer, Luke. BART : Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension. Proceedings of the 58th Annual Meeting of the ACL. 2020

    work page 2020

  49. [49]

    BERT : Pre-training of Deep Bidirectional Transformers for Language Understanding

    Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina. BERT : Pre-training of Deep Bidirectional Transformers for Language Understanding. NAACL. 2019

    work page 2019

  50. [50]

    , booktitle =

    Tom Brown et al. , booktitle =. Language Models are Few-Shot Learners , year =

    work page

  51. [51]

    Communications of the ACM , year=

    Datasheets for datasets , author=. Communications of the ACM , year=

    work page

  52. [52]

    Centre for the Governance of AI

    A guide to writing the NeurIPS impact statement , author=. Centre for the Governance of AI. URL: https://perma. cc/B5R8-2B9V , year=

    work page

  53. [53]

    Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics: Tutorial Abstracts , pages=

    Understanding Ethics in NLP Authoring and Reviewing , author=. Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics: Tutorial Abstracts , pages=

    work page

  54. [54]

    NeurIPS 2021 Paper Checklist Guidelines

    NeurIPS. NeurIPS 2021 Paper Checklist Guidelines

    work page 2021

  55. [55]

    The FAIR Data principles

    FORCE11. The FAIR Data principles

    work page

  56. [56]

    Getting time offsets of beginning and end of each word · openai/whisper · discussion \#3 , url=

    Openai , note =. Getting time offsets of beginning and end of each word · openai/whisper · discussion \#3 , url=. GitHub , year=

    work page

  57. [57]

    A New Order: The Digital Services Act and Consumer Protection , volume=

    CAUFFMAN, Caroline and GOANTA, Catalina , year=. A New Order: The Digital Services Act and Consumer Protection , volume=. European Journal of Risk Regulation , publisher=

    work page

  58. [58]

    Attention is All you Need , year =

    Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez, Aidan N and Kaiser, ukasz and Polosukhin, Illia , booktitle =. Attention is All you Need , year =

    work page

  59. [59]

    International Journal of Data Science and Analytics , year=

    Raza, Shaina and Ding, Chen , title=. International Journal of Data Science and Analytics , year=

    work page

  60. [60]

    LLaMA: Open and Efficient Foundation Language Models

    Llama: Open and efficient foundation language models , author=. arXiv:2302.13971 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  61. [61]

    2023 , eprint=

    Llama 2: Open Foundation and Fine-Tuned Chat Models , author=. 2023 , eprint=

    work page 2023

  62. [62]

    , title =

    Aakanksha Chowdhery et al. , title =. Journal of Machine Learning Research , year =

    work page

  63. [63]

    SemEval-2021 , year=

    SemEval-2021 task 5: Toxic spans detection , author=. SemEval-2021 , year=

    work page 2021

  64. [64]

    Otala, Jacqueline and Liu, Yu and Matthews, Jeanna , title =

    Madraki, Golshan and Grasso, Isabella and M. Otala, Jacqueline and Liu, Yu and Matthews, Jeanna , title =. 2021 , publisher =

    work page 2021

  65. [65]

    Analyzing the Use of Audio Messages in WhatsApp Groups , year =

    Maros, Alexandre and Almeida, Jussara and Benevenuto, Fabr\'. Analyzing the Use of Audio Messages in WhatsApp Groups , year =

    work page

  66. [66]

    Harvard Kennedy School (HKS) Misinformation Review , year=

    Audio misinformation on WhatsApp: A case study from Lebanon , author=. Harvard Kennedy School (HKS) Misinformation Review , year=

    work page

  67. [67]

    2020 , abstract =

    Hussein, Eslam and Juneja, Prerna and Mitra, Tanushree , title =. 2020 , abstract =

    work page 2020

  68. [68]

    Neural Computing and Applications , year=

    Predicting image credibility in fake news over social media using multi-modal approach , author=. Neural Computing and Applications , year=

    work page

  69. [69]

    Nature Comm

    Shao, Chengcheng and Ciampaglia, Giovanni Luca and Varol, Onur and Yang, Kai-Cheng and Flammini, Alessandro and Menczer, Filippo , title=. Nature Comm. , year=

    work page

  70. [70]

    First monday , year=

    Social bots distort the 2016 US Presidential election online discussion , author=. First monday , year=

    work page 2016

  71. [71]

    Overview of CONSTRAINT 2021 Shared Tasks: Detecting English COVID-19 Fake News and Hindi Hostile Posts

    Patwa, Parth and Bhardwaj, Mohit and Guptha, Vineeth and Kumari, Gitanjali and Sharma, Shivam and PYKL, Srinivas and Das, Amitava and Ekbal, Asif and Akhtar, Md Shad and Chakraborty, Tanmoy. Overview of CONSTRAINT 2021 Shared Tasks: Detecting English COVID-19 Fake News and Hindi Hostile Posts. Combating Online Hostile Posts in Regional Languages during Em...

    work page 2021

  72. [72]

    Reis, Julio C. S. and Melo, Philipe and Garimella, Kiran and Almeida, Jussara M. and Eckles, Dean and Benevenuto, Fabrício , journal=

    work page

  73. [73]

    Tardaguila, Cristina and Benevenuto, Fabricio and Ortellado, Pablo , title =

    work page

  74. [74]

    2023 , booktitle =

    Hoseini, Mohamad and Melo, Philipe and Benevenuto, Fabricio and Feldmann, Anja and Zannettou, Savvas , title =. 2023 , booktitle =

    work page 2023

  75. [75]

    First Monday , year=

    QAnon and the information dark age , author=. First Monday , year=

    work page

  76. [76]

    Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

    Reimers, Nils and Gurevych, Iryna. Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks. EMNLP. 2019

    work page 2019

  77. [77]

    ICWSM'19 , year=

    Whatsapp monitor: A fact-checking system for whatsapp , author=. ICWSM'19 , year=

    work page

  78. [78]

    WWW'19 , year=

    (Mis) information dissemination in WhatsApp: Gathering, analyzing and countermeasures , author=. WWW'19 , year=

    work page

  79. [79]

    Telegram Monitor: Monitoring Brazilian Political Groups and Channels on Telegram , year =

    J\'. Telegram Monitor: Monitoring Brazilian Political Groups and Channels on Telegram , year =

    work page