Pith. sign in

REVIEW 4 major objections 5 minor 70 references

Language of Persuasion and Misrepresentation in Business Communication: A Textual Detection Approach

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Pith's one-line read The paper claims that deep-learning classifiers, especially a custom attention model, can sort business texts into factual, persuasive, and misleading categories with validation accuracy up to 97.6%.

desk verdict Abstract overclaims 99% accuracy; best validation is 97.6%, with no independent test set and no released data — a useful cautionary example but not a reliable research contribution. read the letter →

arxiv 2508.09935 v1 pith:VT6EEMXN submitted 2025-08-13 cs.CL q-fin.CPq-fin.GN

classification cs.CLq-fin.CPq-fin.GN
keywords businesscommunicationdeceptiondetectionpersuasionmisrepresentationdeeplearningattentionmechanismtransformertextclassification
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 sets out to establish that persuasion and misrepresentation in business communication leave measurable linguistic traces that automated text classifiers can exploit. To test this, the authors assembled a manually annotated corpus of 4,848 business texts spanning marketing emails, social media, financial statements, and press material, and benchmarked five neural architectures. Their central reported result is that a custom attention-enhanced bidirectional LSTM reaches 97.6% validation accuracy in assigning texts to Factual, Persuasive, or Misleading categories. The abstract states a stronger figure—detection accuracies greater than 99% in controlled settings—than the validation numbers reported in the results. If the accuracy holds outside the lab, companies, auditors, and regulators would gain a scalable way to screen corporate language for deceptive drift.

What carries the argument

The load-bearing object is the custom attention layer attached to a stacked bidirectional LSTM: it computes context-conditioned weights over each token so the model can focus on the specific lexical cues associated with deception—uncertainty markers, exaggerated positive affect, non-immediacy, and framing structures. The supporting machinery is the annotated 4,848-document corpus (1,980 factual, 1,479 persuasive, 1,389 misleading) drawn from 13 text modalities; the corpus defines the three classes, and the attention layer is what the paper credits for the best accuracy and for making predictions interpretable.

What would settle it

Apply the trained Custom Attention model to a newly collected, independently labeled sample of roughly 1,000 business communications from sources not in the training corpus and compare predictions to the new labels. If accuracy falls substantially below 97.6%, or if no evaluation split ever reproduces the abstract's greater-than-99% figure, the paper's central claim is weakened.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that attention-based deep learning can separate business communication into three rhetorical classes with high validation accuracy. The custom attention model—two bidirectional LSTM layers plus a context-weighted attention layer—reaches 0.976 accuracy and is the best of five architectures, ahead of a stacked BiLSTM, a Transformer, and a CNN. The paper argues that attention best captures the deception-relevant cues (uncertainty, extreme positive affect, non-immediacy, framing) that its annotators used, and that the remaining confusion concentrates at the Persuasive–Misleading boundary, which it reads as evidence that the two categories form a linguis

Load-bearing premise

Everything rests on the manual labels: if the five annotators' assignments of Factual, Persuasive, and Misleading are inconsistent, or if the 4,848-text corpus does not represent business communication at large, the reported accuracy numbers will not hold in real-world use.

Editorial extensions

If this is right

  • If the 97.6% validation accuracy generalizes, firms could screen their own outgoing communications before publication as a self-monitoring check.
  • Regulators and auditors could monitor ESG and sustainability claims at scale, connecting to the paper's cited 35% rise in greenwashing.
  • The persistent Persuasive-to-Misleading confusion implies practical deployments should report a deception risk score rather than a hard three-way label.
  • High scores on factual messages suggest the easier half of the problem is identifying facts; the harder half is drawing the line between legitimate persuasion and misrepresentation.
  • The English-only, 4,848-sample corpus means multilingual and larger datasets are needed before deployment, a limitation the paper itself acknowledges.

Reading between the lines

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

  • The abstract's 'greater than 99%' claim does not match the validation numbers reported in the results; the 97.6% Custom Attention figure is the highest accuracy actually shown.
  • Because the corpus labels come from a small expert team, a fresh out-of-domain test set with independent annotation would be the cleanest check on whether the accuracy transfers; the paper does not report one.
  • The paper's continuum reading of Persuasive and Misleading suggests a testable extension: train an ordinal or regression model on a single deception-intensity scale and see whether the boundary errors shrink.
  • The authors' concern about AI-generated business text points to an obvious next benchmark: run the same models on LLM-written marketing and financial language, which is not in the present corpus.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper develops a deep-learning pipeline for classifying business communication into three categories—Factual, Persuasive, and Misleading. It introduces a 4,848-instance English corpus assembled from 13 business-communication modalities, applies a fixed preprocessing and tokenization pipeline, and benchmarks five neural architectures (Simple and Advanced BiLSTM, Custom Attention, Transformer, and CNN). The abstract claims 'detection accuracies of greater than 99% were achieved,' but the reported validation accuracies in Section 3.3 range from about 0.970 to 0.976, with 0.976 as the maximum (Custom Attention, Figure 5). The evaluation uses a single 80/20 train/validation split, with early stopping, learning-rate selection, and checkpoint restoration all performed on the validation split; no independent test set is used. No code, data, or reproducibility details are provided. The paper also includes a broad literature review and discussion of theoretical and practical implications.

Significance. If the reported performance were supported by a rigorous out-of-sample evaluation, the paper would provide a useful multi-architecture benchmark for an understudied domain and would offer empirical grounding for a rhetorical-linguistic taxonomy of business communication. The construction of a multi-source corpus spanning 13 modalities is a genuine strength, as is the systematic comparison of five architectures. However, the central quantitative claim—'greater than 99% accuracy'—is not supported by the paper's own results, and the validation protocol makes even the lower reported accuracies optimistic. The lack of inter-annotator agreement measures, absence of code/data, and incomplete reproducibility information further weaken the contribution. The theoretical synthesis is interesting but secondary; as it stands, the paper's main empirical claim is not credible.

major comments (4)
  1. [Abstract; §3.3, Figures 3–7] The abstract states that 'detection accuracies of greater than 99% were achieved.' The reported validation accuracies never exceed 97.6% (Custom Attention, Figure 5); all other models plateau around 0.970–0.975 (Figures 3, 4, 6, 7). Values near 0.99 appear only in training-accuracy curves, which reflect fit to the training data rather than generalization. Thus the paper's headline quantitative claim is internally inconsistent with its own experimental evidence. This is load-bearing because the 'greater than 99%' figure is the primary basis on which a reader would credit the method's practical utility.
  2. [§3.4] The evaluation protocol uses a single 80/20 train/validation split (Section 3.2). All hyperparameters—including learning rates ('adjusted via initial experiments'), early stopping patience, and model checkpointing ('weights of the best model would be restored based on its validation accuracy')—are selected using the validation split. Consequently, the reported validation accuracies are optimistically biased model-selection results, not out-of-sample estimates. No independent test set or cross-validation is used. To support any accuracy claim, a held-out test set that is not used for early stopping or model selection is required.
  3. [§3.1, Table 2] The corpus is 4,848 documents annotated by 'three business communication researchers, a linguist, and a computational linguist' with third-expert adjudication, but no inter-annotator agreement metric (e.g., Cohen's κ or Krippendorff's α) is reported. Without reliability measures, the subjective labels Factual/Persuasive/Misleading cannot be established as a gold standard. The paper also calls the distribution 'balanced,' yet Table 2 shows 40.8% Factual, 30.5% Persuasive, and 28.7% Misleading; the classes are not equal. This imbalance, while moderate, affects the interpretation of accuracy and confusion matrices.
  4. [§5.4 and overall] The manuscript does not provide code, data, or training seeds, despite Section 5.4 claiming 'the openness of the implementation explanation.' Section 5.4 lists the software environment (Kaggle GPU, TensorFlow, Python 3.8) but omits details essential for reproducibility: random seeds, exact tokenizer configuration beyond vocabulary size, the specific splitting indices, and early-stopping callback logic. The quantitative claims are therefore not independently verifiable. For a journal submission, availability of artifacts or a detailed reproducibility appendix is expected.
minor comments (5)
  1. [§3.1] The sentence 'The resulting dataset is made up of 1,980 Factual, 30.5 (mean: Persuasive contents, and 28.7 Misleading communications' is garbled. It should likely read '1,980 Factual, 1,479 Persuasive, and 1,389 Misleading communications' to match Table 2.
  2. [Figure 9 discussion] The confusion-matrix numbers in the text do not sum consistently. For the Simple LSTM, correct predictions are 378+268+296 = 942 on a 970-sample validation set, leaving 28 errors, but the text lists 3+14+4+11 = 32 errors. Please verify all confusion-matrix arithmetic and the corresponding counts.
  3. [References] Several references are duplicated or misnumbered (e.g., Yang et al. appears as [16] and [18]; Bello et al. appears as [21] and [63]; Park et al. appears as [2] and again as [2] in the literature review). Please clean the reference list and in-text citations.
  4. [Section 4 (implied)] The manuscript jumps from Section 3.4 directly into results discussion without a 'Results' section heading. Add a clear 'Results' or 'Experiments' section header before the discussion beginning 'methodologies in complicated real-world classification problems.'
  5. [Abstract; §3.3.4] The abstract mentions 'personalised transformer models,' but no personalization mechanism is described in Section 3.3.4 or elsewhere. Either define what is meant by 'personalised' or remove the term.

Circularity Check

2 steps flagged · score 6.0 of 10

Headline >99% detection accuracy is a training-fit value; the reported ~97% validation accuracy is measured on the same split used for model selection and early stopping, so the central quantitative claim is an optimistically fitted estimate, not an out-of-sample prediction.

  1. fitted input called prediction [Abstract and Section 3.3.3 (Figure 5)]
    "In controlled settings, detection accuracies of greater than 99% were achieved by using computational textual analysis as well as personalised transformer models. ... the training accuracy (blue) continuing to advance towards around 0.99 at epoch six and the validation accuracy (orange) gathering maximum accuracy of 0.976 at epoch 7."

    The only numbers >0.99 in the manuscript are training-accuracy curves. Training accuracy is the quantity minimized during fitting, so it is a fitted value, not a predictive detection accuracy. The paper's own validation maximum is 0.976 (Custom Attention, Figure 5), so the abstract's headline 'detection accuracies >99%' is a training fit relabeled as a detection result.

  2. fitted input called prediction [Sections 3.2 and 3.4]
    "the corpus was divided into two mutually exclusive subsets, a training and validation set... Each architecture had its learning rates adjusted via initial experiments to find the preferred ratio of training efficiency to stability... Early-stopping was done using validation loss, with a patience value of 3 epochs, and the weights of the best model would be restored based on its validation accuracy. Model checkpointing was used to save the model weights that yielded the highest validation accuracy during training."

    The validation split is used to tune learning rates, to stop training, and to select the best checkpoint. The reported accuracies (0.97-0.976) are measured on that same validation set, so they are the optimized selection criterion, not an independent test-set prediction. With no separate test set, the central quantitative claim is an in-sample fit to the validation data rather than a prediction of unseen business communication.

full rationale

The paper is primarily an empirical benchmark with a literature review; there is no formal derivation chain, and no load-bearing self-citation is present. The circularity is concentrated in the quantitative claim. The abstract's >99% accuracy figure appears nowhere in the results: the body's validation accuracies peak at 0.976, while the ~0.99 values appear only in training-accuracy curves, which are fitting diagnostics by construction. Additionally, the reported validation accuracy was used for early stopping, learning-rate adjustment, and checkpoint selection, so it is not an unbiased out-of-sample performance estimate. The absence of a separate test set or external benchmark means the paper's central 'detection accuracy' number reduces to performance on the data used to fit and select the models. This is a partial circularity in the headline result; the underlying annotated corpus and model comparisons still have independent empirical content, which is why the score is 6 rather than higher. The overlap between annotation cues and the attention model's intended cues is a validity threat but not a formal circularity, since the attention weights are learned.

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

The central claim rests on the quality and representativeness of the private labeled corpus, the assumption that linguistic cues are diagnostic of deception, and the use of validation accuracy as the reported performance. No external benchmarks, public data, or released code support the claim.

free parameters (5)
  • Learning rates per model = 0.001 (Simple LSTM, Advanced LSTM, CNN), 0.0005 (Custom Attention), 0.0001 (Transformer)
    Chosen via initial experiments on the validation split; they affect the final accuracy reported.
  • Maximum sequence length = 100 tokens
    Chosen because 95% of texts fall within it; truncation choice affects input representation.
  • Vocabulary size = 10,000
    Imposed by Keras tokenizer; affects coverage of rare deception cues.
  • Train/validation split = 80/20 (3,878/970 instances)
    Split choice affects variance of validation estimates; no test set is held out.
  • Early stopping patience = 3 epochs
    Model selection uses validation loss; this makes reported validation accuracy an optimistic estimate.
assumptions (4)
  • domain assumption The three classes Factual, Persuasive, and Misleading are mutually exclusive, exhaustive, and reliably annotatable by five experts.
    Section 3.1 treats expert labels as ground truth; if the boundary between persuasion and misleading language is not stable, accuracy numbers are not meaningful.
  • domain assumption Surface linguistic cues (pronouns, metadiscourse, framing, uncertainty, positive affect) reliably carry deception signals.
    The annotation scheme and feature choices in Sections 3.1 and 3.2 assume these cues are diagnostic; the cited literature itself debates this.
  • domain assumption Validation-set accuracy on this private corpus estimates real-world detection performance.
    Section 3.4 and Section 4 use validation accuracy as the headline result; Section 5.5 admits the controlled environment may not reflect operational use.
  • standard math Standard neural text classifiers can achieve high F1 on deception tasks (from cited prior works).
    The paper relies on established deep learning results (references 22, 27 to 29) as motivation; these are treated as background rather than claims needing proof here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Language of Persuasion and Misrepresentation in Business Communication: A Textual Detection Approach." pith.science (2026). https://pith.science/paper/VT6EEMXN

@misc{pith2026250809935,
  author       = {Pith},
  title        = {Pith review of: Language of Persuasion and Misrepresentation in Business Communication: A Textual Detection Approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VT6EEMXN}},
  note         = {Machine review of arXiv:2508.09935}
}
read the original abstract

Business communication digitisation has reorganised the process of persuasive discourse, which allows not only greater transparency but also advanced deception. This inquiry synthesises classical rhetoric and communication psychology with linguistic theory and empirical studies in the financial reporting, sustainability discourse, and digital marketing to explain how deceptive language can be systematically detected using persuasive lexicon. In controlled settings, detection accuracies of greater than 99% were achieved by using computational textual analysis as well as personalised transformer models. However, reproducing this performance in multilingual settings is also problematic and, to a large extent, this is because it is not easy to find sufficient data, and because few multilingual text-processing infrastructures are in place. This evidence shows that there has been an increasing gap between the theoretical representations of communication and those empirically approximated, and therefore, there is a need to have strong automatic text-identification systems where AI-based discourse is becoming more realistic in communicating with humans.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 68 canonical work pages

  1. [1]

    A statistical language modeling approach to online deception detection

    Lina Zhou, Yongmei Shi, and Dongsong Zhang. A statistical language modeling approach to online deception detection. IEEE Transactions on Knowledge and Data Engineering , 20(8):1077--1091, 2008

  2. [2]

    Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks

    Peter S. Park, Simon Goldstein, Aidan O’Gara, Michael Chen, and Dan Hendrycks. Ai deception: A survey of examples, risks, and potential solutions. Patterns , 5(5):100988, 2024

  3. [3]

    Textual analysis in accounting: What's next? Contemporary Accounting Research , 40(2):765--805, 2023

    Khrystyna Bochkay, Stephen V Brown, Andrew J Leone, and Jennifer Wu Tucker. Textual analysis in accounting: What's next? Contemporary Accounting Research , 40(2):765--805, 2023

  4. [4]

    Identification of fraudulent financial statements using linguistic credibility analysis

    Sean L Humpherys, Kevin C Moffitt, Mary B Burns, Judee K Burgoon, and William F Felix. Identification of fraudulent financial statements using linguistic credibility analysis. Decision Support Systems , 50(3):585--594, 2011

  5. [5]

    Enhancing environmental information transparency through corporate social responsibility reporting regulation

    Fabio Caputo, Simone Pizzi, Lorenzo Ligorio, and Rossella Leopizzi. Enhancing environmental information transparency through corporate social responsibility reporting regulation. Business Strategy and the Environment , 30(8):3470--3484, 2021

  6. [6]

    Mapping the greenwashing research landscape: A theoretical and field analysis

    Canio Forliano, Enrico Battisti, Paola de Bernardi, and Tomáš Kliestik. Mapping the greenwashing research landscape: A theoretical and field analysis. Review of Managerial Science , 2025

  7. [7]

    Detecting and unmasking ai-generated texts through explainable artificial intelligence using stylistic features

    Aditya Shah, Prateek Ranka, Urmi Dedhia, Shruti Prasad, Siddhi Muni, and Kiran Bhowmick. Detecting and unmasking ai-generated texts through explainable artificial intelligence using stylistic features. International Journal of Advanced Computer Science and Applications , 14(10), 2023

  8. [8]

    Carillion's strategic choices and the boardroom's strategies of persuasive appeals: ethos, logos and pathos

    Fadi Alkaraan, Mohammad Albahloul, and Khaled Hussainey. Carillion's strategic choices and the boardroom's strategies of persuasive appeals: ethos, logos and pathos. Journal of Applied Accounting Research , 24, 01 2023

Show all 70 references
  1. [9]

    Walking the talk about corporate social responsibility communication: An elaboration likelihood model perspective

    Mark Anthony Camilleri. Walking the talk about corporate social responsibility communication: An elaboration likelihood model perspective. Business Ethics, the Environment & Responsibility , 2022

  2. [10]

    Flusberg, Kevin J

    Stephen J. Flusberg, Kevin J. Holmes, Paul H. Thibodeau, Robin L. Nabi, and Teenie Matlock. The psychology of framing: How everyday language shapes the way we think, feel, and act. Psychological Science in the Public Interest , 25(3):105--161, 2024

  3. [11]

    Sustainable finance as a contested concept: Tracing the evolution of five frames between 1998 and 2018

    Andreas Dimmelmeier. Sustainable finance as a contested concept: Tracing the evolution of five frames between 1998 and 2018. Journal of Sustainable Finance & Investment , 2021

  4. [12]

    Elena N. Malyuga. A corpus-based approach to corporate communication research. Russian Journal of Linguistics , 27(1):152--172, 2023

  5. [13]

    Using metadiscourse to enhance persuasiveness in corporate press releases: A corpus-based study

    Shuai Liu and Jingyuan Zhang. Using metadiscourse to enhance persuasiveness in corporate press releases: A corpus-based study. SAGE Open , 11(3):1--11, 2021

  6. [14]

    Anthropomorphization and beyond: conceptualizing humanwashing of ai-enabled machines

    Gabriela Scorici, Mario D Schultz, and Peter Seele. Anthropomorphization and beyond: conceptualizing humanwashing of ai-enabled machines. AI & Society , 39:789--795, 2024

  7. [15]

    Voluntary disclosure of sustainable development goals in mandatory non-financial reports: The moderating role of cultural dimension

    Simone Pizzi, Mara Del Baldo, Fabio Caputo, and Andrea Venturelli. Voluntary disclosure of sustainable development goals in mandatory non-financial reports: The moderating role of cultural dimension. Journal of International Financial Management & Accounting , 2021

  8. [16]

    Makana Chock

    Jeongwon Yang, Ploypin Chuenterawong, Heejae Lee, Yu Tian, and T. Makana Chock. Human versus virtual influencer: The effect of humanness and interactivity on persuasive csr messaging. Journal of Interactive Advertising , 23(1):1--15, 2023

  9. [17]

    Ethical and Legal Challenges of AI in Marketing : An Exploration of Solutions

    Dinesh Kumar. Ethical and Legal Challenges of AI in Marketing : An Exploration of Solutions . SSRN Electronic Journal , 2023

  10. [18]

    Finchain-bert: A high-accuracy automatic fraud detection model based on nlp methods for financial scenarios

    Xinze Yang, Chunkai Zhang, Yizhi Sun, Kairui Pang, Luru Jing, Shiyun Wa, and Chunli Lv. Finchain-bert: A high-accuracy automatic fraud detection model based on nlp methods for financial scenarios. Information , 14(9):499, 2023

  11. [19]

    How will ai text generation and processing impact sustainability reporting? critical analysis, a conceptual framework, and avenues for future research

    Charl De Villiers, Ruth Dimes, and Matteo Molinari. How will ai text generation and processing impact sustainability reporting? critical analysis, a conceptual framework, and avenues for future research. Sustainability Accounting, Management and Policy Journal , 2023

  12. [20]

    Corporate sustainability communication as ‘fake news’: Firms’ greenwashing on twitter

    Divinus Oppong-Tawiah and Jane Webster. Corporate sustainability communication as ‘fake news’: Firms’ greenwashing on twitter. Sustainability , 15(8):6683, 2023

  13. [21]

    Machine learning approaches for enhancing fraud prevention in financial transactions

    Oluwabusayo Adijat Bello, Adebola Folorunso, Oluomachi Eunice Ejiofor, Folake Zainab Budale, Kayode Adebayo, and Olayemi Alex Babatunde. Machine learning approaches for enhancing fraud prevention in financial transactions. International Journal of Management Technology , 10(1)...

  14. [22]

    Fake reviews classification using deep learning ensemble of shallow convolutions

    Muhammad Saad Javed, Hammad Majeed, Hasan Mujtaba, and Mirza Omer Beg. Fake reviews classification using deep learning ensemble of shallow convolutions. Journal of Computational Social Science , 4:1--26, 2021

  15. [23]

    Fraud detection in healthcare insurance claims using machine learning

    Eman Nabrawi and Abdullah Alanazi. Fraud detection in healthcare insurance claims using machine learning. Risks , 11(9):160, 2023

  16. [24]

    Artificial intelligence in finance: A comprehensive review through bibliometric and content analysis

    Salman Bahoo, Marco Cucculelli, Xhoana Goga, and Jasmine Mondolo. Artificial intelligence in finance: A comprehensive review through bibliometric and content analysis. SN Business & Economics , 4(1):23, 2024

  17. [25]

    Anis, Reef M

    Nida Aslam, Irfan Ullah Khan, Samiha Mirza, Alanoud AlOwayed, Fatima M. Anis, Reef M. Aljuaid, and Reham Baageel. Interpretable machine learning models for malicious domains detection using explainable artificial intelligence (xai). Sustainability , 14(12):7375, 2022

  18. [26]

    Attentive statement fraud detection: Distinguishing multimodal financial data with fine-grained attention

    Gang Wang, Jingling Ma, and Gang Chen. Attentive statement fraud detection: Distinguishing multimodal financial data with fine-grained attention. Decision Support Systems , 167:113913, 2023

  19. [27]

    A deep learning method for automatic sms spam classification: Performance of learning algorithms on indigenous dataset

    Olusola Abayomi-Alli, Sanjay Misra, and Adebayo Abayomi-Alli. A deep learning method for automatic sms spam classification: Performance of learning algorithms on indigenous dataset. Concurrency and Computation: Practice and Experience , 34(17):e6989, 2022

  20. [28]

    Oswald, Sona Elza Simon, and Arnab Bhattacharya

    C. Oswald, Sona Elza Simon, and Arnab Bhattacharya. Spotspam: Intention analysis driven sms spam detection using bert embeddings. ACM Transactions on the Web , 16(3):1--24, 2022

  21. [29]

    Advancing fake news detection: Hybrid deep learning with fasttext and explainable ai

    Ehtesham Hashmi, Sule Yildirim Yayilgan, Muhammad Mudassar Yamin, Subhan Ali, and Mohamed Abomhara. Advancing fake news detection: Hybrid deep learning with fasttext and explainable ai. IEEE Access , 12:44462--44482, 2024

  22. [30]

    Intelligent financial fraud detection practices in post-pandemic era

    Xiaoqian Zhu, Xiang Ao, Zidi Qin, Yanpeng Chang, Yang Liu, Qing He, and Jianping Li. Intelligent financial fraud detection practices in post-pandemic era. The Innovation , 2(4):100176, 2021

  23. [31]

    A bert based approach to measure web services policies compliance with gdpr

    Lavanya Elluri, Sai Sree Laya Chukkapalli, Karuna Pandey Joshi, Tim Finin, and Anupam Joshi. A bert based approach to measure web services policies compliance with gdpr. IEEE Access , 9:148004--148015, 2021

  24. [32]

    U. Gupta. Gpt-investar: Enhancing stock investment strategies through annual report analysis with large language models, 2023

  25. [33]

    Extracting financial data from unstructured sources: Leveraging large language models

    Huaxia Li, Haoyun Gao, Chengzhang Wu, and Miklos A Vasarhelyi. Extracting financial data from unstructured sources: Leveraging large language models. Available at SSRN 4567607 , 2023

  26. [34]

    Rethinking legal compliance automation: Opportunities with large language models

    Shabnam Hassani, Mehrdad Sabetzadeh, Daniel Amyot, and Jain Liao. Rethinking legal compliance automation: Opportunities with large language models. arXiv preprint arXiv:2404.14356 , 2024

  27. [35]

    Sniffer: Multimodal large language model for explainable out-of-context misinformation detection

    Peng Qi, Zehong Yan, Wynne Hsu, and Mong Li Lee. Sniffer: Multimodal large language model for explainable out-of-context misinformation detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 13052--13061, 2024

  28. [36]

    DEAP-FAKED: Knowledge Graph based Approach for Fake News Detection

    Mohit Mayank, Shakshi Sharma, and Rajesh Sharma. DEAP-FAKED: Knowledge Graph based Approach for Fake News Detection . In IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) 2022 , 2022

  29. [37]

    Fake review detection in e-commerce platforms using aspect-based sentiment analysis

    Petr Hajek, Lubica Hikkerova, and Jean-Michel Sahut. Fake review detection in e-commerce platforms using aspect-based sentiment analysis. Preprint , 2022

  30. [38]

    From opinion mining to financial argument mining

    Chung-Chi Chen, Hen-Hsen Huang, and Hsin-Hsi Chen. From opinion mining to financial argument mining . SpringerBriefs in Computer Science. Springer, Cham, 2021

  31. [39]

    Analyzing and visualizing text information in corporate sustainability reports using natural language processing methods

    Hyewon Kang and Jinho Kim. Analyzing and visualizing text information in corporate sustainability reports using natural language processing methods. Applied Sciences , 12(11):5614, 2022

  32. [40]

    Three gaps in computational text analysis methods for social sciences: A research agenda

    Christian Baden, Christian Pipal, Martijn Schoonvelde, and Mariken ACG van der Velden. Three gaps in computational text analysis methods for social sciences: A research agenda. Communication Methods and Measures , 16(1):1--18, 2022

  33. [41]

    Missing counter-evidence renders nlp fact-checking unrealistic for misinformation

    Max Glockner, Yufang Hou, and Iryna Gurevych. Missing counter-evidence renders nlp fact-checking unrealistic for misinformation. arXiv preprint arXiv:2210.13865 , 2022

  34. [42]

    Larcker and Anastasia A

    David F. Larcker and Anastasia A. Zakolyukina. Detecting deceptive discussions in conference calls. Journal of Accounting Research , 50(2):495--540, 2012

  35. [43]

    Exploring top management language for signals of possible deception: The words of satyam's chair ramalinga raju

    Russell Craig, Tony Mortensen, and Shefali Iyer. Exploring top management language for signals of possible deception: The words of satyam's chair ramalinga raju. Journal of Business Ethics , 113(3):333--347, 2013

  36. [44]

    Burgoon, Douglas P

    Lina Zhou, Judee K. Burgoon, Douglas P. Twitchell, Tiantian Qin, and Jay F. Nunamaker. A comparison of classification methods for predicting deception in computer-mediated communication. Journal of Management Information Systems , 20(4):139--165, 2004

  37. [45]

    Accounting variables, deception, and a bag of words: Assessing the tools of fraud detection

    Lynnette Purda and David Skillicorn. Accounting variables, deception, and a bag of words: Assessing the tools of fraud detection. Journal of Forensic & Investigative Accounting , 4(1):1--25, 2012

  38. [46]

    Deceptive opinion spam detection using neural network

    Yafeng Ren and Yue Zhang. Deceptive opinion spam detection using neural network. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers , pages 140--150. Osaka, Japan, 2016

  39. [47]

    Pay attention and you won’t lose it: A deep learning approach to sequence imputation

    Ilya Sucholutsky, Akshay Narayan, Matthias Schonlau, and Sebastian Fischmeister. Pay attention and you won’t lose it: A deep learning approach to sequence imputation. PeerJ Computer Science , 5:e210, 2019

  40. [48]

    Vickers, L

    P. Vickers, L. Barrault, E. Monti, and N. Aletras. We need to talk about classification evaluation metrics in nlp. In Proceedings of the 13th International Joint Conference on Natural Language Processing (IJCNLP 2023) , volume 1, pages 498--510, 2023

  41. [49]

    Yoo, Chan Yeob Yeun, Dirar Homouz, and Ahmed Taha

    Khaled Taha, Paul D. Yoo, Chan Yeob Yeun, Dirar Homouz, and Ahmed Taha. A comprehensive survey of text classification techniques and their research applications: Observational and experimental insights. Computer Science Review , 54:100664, 2024

  42. [50]

    F1 score in machine learning explained

    Encord . F1 score in machine learning explained. https://encord.com/blog/f1-score-in-machine-learning/, 2024. Accessed: 2025-08-06

  43. [51]

    Classification: Accuracy, recall, precision, and related metrics

    Google Developers . Classification: Accuracy, recall, precision, and related metrics. https://developers.google.com/machine-learning/crash-course/classification/accuracy-precision-recall, 2023. Accessed: 2025-08-06

  44. [52]

    Understanding and applying f1 score: Ai evaluation essentials

    Arize . Understanding and applying f1 score: Ai evaluation essentials. https://arize.com/blog-course/f1-score/, 2024. Accessed: 2025-08-06

  45. [53]

    Custom text classification evaluation metrics

    Azure AI . Custom text classification evaluation metrics. https://learn.microsoft.com/en-us/azure/ai-services/language-service/custom-text-classification/concepts/evaluation-metrics, 2024. Accessed: 2025-08-06

  46. [54]

    S. A. Prome, N. A. Ragavan, M. R. Islam, D. Asirvatham, and A. J. Jegathesan. Deception detection using machine learning (ml) and deep learning (dl) techniques: A systematic review. Natural Language Processing Journal , 6:100057, 2024

  47. [55]

    Decoding persuasion: a survey on ml and nlp methods for the study of online persuasion

    Davide Bassi, Søren Fomsgaard, and Martín Pereira-Fariña. Decoding persuasion: a survey on ml and nlp methods for the study of online persuasion. Frontiers in Communication , 9:1457433, 2024

  48. [56]

    Mohawesh, H

    R. Mohawesh, H. Bany Salameh, Y. Jararweh, M. Alkhalaileh, and S. Maqsood. Fake review detection using transformer-based enhanced lstm and roberta. International Journal of Cognitive Computing in Engineering , 5:250--258, 2024

  49. [57]

    A comprehensive analysis of deception detection techniques leveraging machine learning

    Hagar Elbatanouny, Noora Al Roken , Abir Hussain, Wasiq Khan, Bilal Khan, and Eqab Almajali. A comprehensive analysis of deception detection techniques leveraging machine learning. Expert Systems with Applications , 283:127601, 2025

  50. [58]

    Swaminathan and B

    S. Swaminathan and B. R. Tantri. Confusion matrix-based performance evaluation metrics. African Journal of Biomedical Research , 27(4s):4023--4031, 2024

  51. [59]

    S. L. Humpherys, K. C. Moffitt, M. B. Burns, J. K. Burgoon, and W. F. Felix. Identification of fraudulent financial statements using linguistic credibility analysis. Decision Support Systems , 50(3):587--594, 2011

  52. [60]

    A systematic review of aspect-based sentiment analysis: domains, methods, and trends

    Yan Cathy Hua, Paul Denny, J \"o rg Wicker, and Katerina Taskova. A systematic review of aspect-based sentiment analysis: domains, methods, and trends. Artificial Intelligence Review , 57:296, 2024

  53. [61]

    Environmental claim detection, 2023

    Dominik Stammbach, Nicolas Webersinke, Julia Anna Bingler, Mathias Kraus, and Markus Leippold. Environmental claim detection, 2023

  54. [62]

    The effects of communication media and culture on deception detection accuracy

    Joey F George, Manjul Gupta, Gabriel Giordano, Annette M Mills, Vanesa M Tennant, and Carmen C Lewis. The effects of communication media and culture on deception detection accuracy. MIS Quarterly , 42(2):513--542, 2018

  55. [63]

    Ai-driven approaches for real-time fraud detection in us financial transactions: Challenges and opportunities

    Oluwabusayo Adijat Bello, Abidemi Ogundipe, Damilola Mohammed, Adebola Folorunso, and Olalekan Ayodeji Alonge. Ai-driven approaches for real-time fraud detection in us financial transactions: Challenges and opportunities. European Journal of Computer Science and Information Te...

  56. [64]

    Albuquerque

    Akshi Kumar, Shubham Dikshit, and Victor Hugo C. Albuquerque. Explainable artificial intelligence for sarcasm detection in dialogues. Wireless Communications and Mobile Computing , 2021:1--13, 2021

  57. [65]

    Fake news detection: a survey of evaluation datasets

    Arianna D'Ulizia, Maria Chiara Caschera, Fernando Ferri, and Patrizia Grifoni. Fake news detection: a survey of evaluation datasets. PeerJ Computer Science , 7:e518, 2021

  58. [66]

    Financial fraud detection using vocal, linguistic and financial cues

    Chandra S Throckmorton, William J Mayew, Mohan Venkatachalam, and Leslie M Collins. Financial fraud detection using vocal, linguistic and financial cues. Decision Support Systems , 74:78--87, 2015

  59. [67]

    Hossen and M

    S. Hossen and M. M. Joti. Dataset for: Language of persuasion and misrepresentation in business communication: A textual detection approach (version 1) [data set], 2025

  60. [68]

    Hossen and M

    S. Hossen and M. M. Joti. Figures for: Language of persuasion and misrepresentation in business communication: A textual detection approach (version 1), 2025

  61. [69]

    Hossen and M

    S. Hossen and M. M. Joti. Models of: Language of persuasion and misrepresentation in business communication: A textual detection approach, 2025

  62. [70]

    Hossen and M

    S. Hossen and M. M. Joti. Source code for: Language of persuasion and misrepresentation in business communication: A textual detection approach, 2025

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.