Semantics of Subterfuge: Benchmarking Legal Deception Detection Against General-domain State-of-the-Art

# Semantics of Subterfuge: Benchmarking Legal Deception Detection Against General-domain State-of-the-Art
Source: [https://arxiv.org/html/2607.29066](https://arxiv.org/html/2607.29066)
\(2026\)

###### Abstract\.

Deception detection has critical implications for legal proceedings, law enforcement, and online security\. Although human judgment is limited in accuracy and scalability, Natural Language Processing \(NLP\) offers a data\-driven alternative\. We present a survey and comparative analysis of NLP\-based Automatic Deception Detection \(ADD\) focusing on the legal domain, reviewing the evolution from feature\-based machine learning to Large Language Model \(LLM\) approaches\. We conduct a unified empirical evaluation across seven datasets \(two legal, five general\-domain\), comparing six fine\-tuned transformer models and seven LLMs under four prompting strategies\. The results show strong domain sensitivity, with fine\-tuned models excelling in data\-rich general domains and few\-shot LLMs remaining competitive in low\-resource legal settings\. Chain\-of\-Thought prompting often underperforms direct classification\. These findings highlight the need for domain adaptation and interpretable systems in high\-stakes legal contexts\.

Deception detection, Natural Language Processing, Machine Learning, Large Language Models, Legal deception, Linguistic analysis

††copyright:acmlicensed††journalyear:2026††doi:XXXXXXX\.XXXXXXX††ccs:Computing methodologies Natural language processing††ccs:Applied computing Law††ccs:Computing methodologies Machine learning## 1\.Introduction

Deception detection is a critical challenge in various domains, including criminal investigations, legal proceedings, fraud prevention, and social media, with major implications for law enforcement and judicial systems\. In legal contexts, the stakes are particularly high: undetected deception can obstruct justice, while false positives risk wrongful accusations and loss of public trust\. Current practises are based on subjective judgment and resource\-intensive manual analysis of statements and testimony\. Automated deception detection systems could support decision\-making by flagging suspicious statements while reducing the cognitive load on human analysts\. However, the legal domain poses distinct challenges: limited training data due to privacy constraints, high costs of false positives, and strict requirements for interpretability and accountability\.

### 1\.1\.Scope and Contributions

We present a structured review and comparative analysis of NLP\-based deception detection with emphasis on legal and police contexts\. We trace the evolution from traditional machine learning to modern LLM approaches, and conduct unified experiments across seven public datasets \(two legal, five general\-domain\), comparing six fine\-tuned transformer models \(RoBERTa, BERT, DeBERTa, ALBERT, DistilBERT, T5\) and seven LLMs \(GPT\-4o, LLaMA, Gemma2, Phi variants\) under four prompting strategies\. Our analysis reveals domain sensitivity patterns, prompting strategy effectiveness, temperature sensitivity, and practical limitations to guide future research toward effective and accountable ADD systems\.

## 2\.Related Work

### 2\.1\.Evolution of NLP in Deception Detection

Early deception research primarily focused on non\-verbal signals\. However, increasing evidence suggests that linguistic behaviour provides more reliable and scalable indicators of deceptive intent, enabling computational analysis without continuous human intervention\. Empirical studies indicate that human judges perform only slightly above chance \(54\-60% accuracy\) in deception detection\(Liet al\.,[2020](https://arxiv.org/html/2607.29066#bib.bib1)\), motivating the adoption of computational approaches that can identify subtle linguistic patterns across large text corpora\.

Text\-based Automatic Deception Detection \(ADD\) has evolved through three main phases\. Early methods relied on manually engineered linguistic and psychological features \(Linguistic Inquiry and Word Count \(LIWC\) indicators, n\-grams, POS \(Part\-of\-Speech\) tags\) combined with classical machine learning classifiers \(Support Vector Machines \(SVM\), Random Forests\)\. Neural representation learning enabled models to capture semantic and contextual information directly from text, reducing dependence on handcrafted features\. Most recently, large language models \(LLMs\) have introduced knowledge\-augmented approaches leveraging pre\-training, prompt engineering, few\-shot learning, and parameter\-efficient fine\-tuning for context\-sensitive reasoning about deceptive language\(Zhang and Gao,[2023](https://arxiv.org/html/2607.29066#bib.bib33)\)\.

### 2\.2\.Domain\-Specific Applications: Legal and Policing

In legal contexts, deception detection supports trial transcript analysis, evidence validation, and credibility assessment, which raises ethical concerns around interpretability and fairness\. Police\-oriented research addresses complaint verification, misconduct detection, and false report identification\. Table[1](https://arxiv.org/html/2607.29066#S2.T1)summarises key work in both domains\. Challenges include data scarcity, sensitivity, and high misclassification costs; thus, human oversight remains essential for responsible deployment\.

Table 1\.Work Related to the Legal and Police Domain

## 3\.Comparative Analysis

### 3\.1\.Experimental Setup

#### 3\.1\.1\.DataSets

Based on the existing work, we selected seven datasets: two legal domain datasets and five general domain datasets with the aim of evaluating cross\-domain model behaviour\. Table[2](https://arxiv.org/html/2607.29066#S3.T2)summarises corpus statistics\.

Table 2\.Dataset Statistical Information\.SC: average Sentence Count,AT: Average Token count,VR: Vocabulary Richness##### Real\-Life Trial Deception \(RLTD\)\(Pérez\-Rosaset al\.,[2015](https://arxiv.org/html/2607.29066#bib.bib32)\)

was created using the videos collected from public court trials in USA\. The deception labels \(truthful/deceptive\) were determined by the trial outcomes\. The videos were then transcribed via crowdsourcing to capture verbal information, and manually annotated for non\-verbal cues such as facial displays and hand movements\. For our experiments, we used the transcribed text without the non\-verbal cues\.

##### DECOUR\(Fornaciariet al\.,[2012](https://arxiv.org/html/2607.29066#bib.bib105)\)

includes Italian courtroom transcripts from 35 criminal hearings, comprising dialogues between interviewees and interviewers \(judges, prosecutors, lawyers\)\. Utterances were labelled True, False, or Uncertain\. For our experiments, only True/False utterances were used to maintain binary classification consistency across datasets\.

##### General Domain Datasets

OpSpam\(Ottet al\.,[2011](https://arxiv.org/html/2607.29066#bib.bib109)\)consists of hotel reviews collected fromTripAdvisorcovering 20 Chicago hotels\. Cross\-cultural Deception Detection\(cCult\)\(Pérez\-Rosas and Mihalcea,[2014](https://arxiv.org/html/2607.29066#bib.bib44)\)dataset was collected via crowdsourcing with truthful and deceptive short essays on opinions of three topics \(Abortion, Death Penalty, Best Friend\)\.DeRev2014\(Fornaciari and Poesio,[2014](https://arxiv.org/html/2607.29066#bib.bib101)\)is a corpus of 236 book reviews,Liar\(Wang,[2017](https://arxiv.org/html/2607.29066#bib.bib104)\)andFakeNewsNet\(Shuet al\.,[2020](https://arxiv.org/html/2607.29066#bib.bib106)\)are fake news datasets sourced fromPolitiFactandBuzzFeed\.

Table[2](https://arxiv.org/html/2607.29066#S3.T2)reports corpus sizes, label distributions, and linguistic characteristics\. SC and token counts were computed using NLTK; vocabulary richness \(VR\) is defined as the type\-token ratio, where lower values indicate more repetitive language\.

#### 3\.1\.2\.Data processing

We applied unified 80:10:10 train\-validation\-test splits across all datasets, following original split ratios of Liar\(Wang,[2017](https://arxiv.org/html/2607.29066#bib.bib104)\)\. However, its original six labels were mapped to binary: true/mostly\-true/half\-true→\\toTrue and false/barely\-true/pants\-on\-fire→\\toFalse\.

#### 3\.1\.3\.Models

We evaluated encoder\-only transformers \(RoBERTa, BERT, DeBERTa, ALBERT, DistilBERT\) and T5\-base via supervised fine\-tuning, alongside seven LLMs: GPT\-4o, GPT\-4o\-mini, LLaMA3\-8B, LLaMA3\.1\-8B, Gemma2\-9B, Phi\-3\-mini, and Phi\-4, which have been used in recent work on deception detection\(Benny,[2023](https://arxiv.org/html/2607.29066#bib.bib107); Velutharambath and Klinger,[2023](https://arxiv.org/html/2607.29066#bib.bib99); Miahet al\.,[2025](https://arxiv.org/html/2607.29066#bib.bib21); Cuiet al\.,[2025](https://arxiv.org/html/2607.29066#bib.bib91); Aspromonteet al\.,[2025](https://arxiv.org/html/2607.29066#bib.bib87); Papantoniouet al\.,[2025](https://arxiv.org/html/2607.29066#bib.bib56)\)\. Weighted F1 is used throughout to account for class imbalance\.

#### 3\.1\.4\.Experimental Setup

Fine\-tuning used: learning rate=2e\-5, batch size=8, epochs=6, AdamW optimiser, on an NVIDIA RTX 3080 \(16GB\)\. Open\-source LLMs were run locally via Ollama \(v0\.13\.5\) with 4\-bit quantisation \(Q4\_K\_M\)\. OpenAI models were accessed via the Chat Completions API\. All LLMs used temperature=0 for deterministic outputs unless stated otherwise\.

### 3\.2\.Prompting Strategies

#### 3\.2\.1\.Prompt Configurations

We evaluated five distinct experimental configurations to assess the impact of fine\-tuning and prompting strategies: \(1\)Supervised Fine\-tuning:BERT\-based models and T5\-base models were fine\-tuned on training data using standard cross\-entropy loss with the hyperparameters specified in Section 3\.1\.4\. \(2\)Zero\-shot Direct Classification:LLMs received a task description and text input without examples: \(3\)Few\-shot Direct Classification:Four selected examples from the training set were prepended: \(4\)Zero\-shot Chain\-of\-Thought \(CoT\):Models were instructed to reason before classifying: \(5\)Few\-shot Chain\-of\-Thought \(CoT\):Combining examples with reasoning demonstrations \(4 examples with step\-by\-step reasoning\)\.

#### 3\.2\.2\.Few\-shot Example Selection

For all few\-shot configurations,k=4k\{=\}4examples were selected dynamically per test instance using sentence embeddings \(all\-MiniLM\-L6\-v2\), with label balancing enforced \(2 truthful, 2 deceptive\)\.Top\-Kselects thekkmost similar examples to the query by cosine similarity\.High\-Varianceuses a greedy procedure seeded by the most similar example, iteratively adding candidates that maximise pairwise similarity variance within the selected set, producing a linguistically diverse context\.

## 4\.Results and Observations

Fine\-tuned transformers and LLMs were evaluated under consistent splits and prompts with no dataset\-specific tuning\. Full results are shown in Table[3](https://arxiv.org/html/2607.29066#S4.T3)\.

Table 3\.F1 Scores \(%\) for Deception Detection Across Datasets and Models\. FT=Fine\-tuned; ZS=Zero\-shot; TK=Few\-shot Top\-K; HV=Few\-shot High\-Variance; CoT=Chain\-of\-Thought\. Bold indicates best per dataset per section\.Table 4\.F1 Weighted Scores \(%\) — 10\-Fold CV vs Original \(RLTD\)Table 5\.F1 Weighted Scores \(%\) — LLM Direct Classification, 10\-Fold CV Across Temperatures \(RLTD\)##### Fine\-tuned Models vs\. Zero\-shot LLMs

Fine\-tuned transformer models consistently outperformed zero\-shot LLMs on datasets with sufficient data\. DeBERTa achieves F1 = 92\.52% on OpSpam versus GPT\-4o’s 59\.63%; multiple fine\-tuned models reach 100% on DeRev2014 hough this reflects dataset artefacts \(see Section[5](https://arxiv.org/html/2607.29066#S5)\)\. However, on the small RLTD dataset \(121 samples\), few\-shot GPT\-4o \(F1=84\.62%\) matches or exceeds fine\-tuned models, suggesting pre\-trained priors compensate when labelled data is scarce\.

##### Domain Sensitivity

Legal datasets yield lower, more variable performance: RLTD spans 14\.48–84\.62% across models; DECOUR peaks at 71\.95% \(DistilBERT\)\. Review datasets are most tractable \(OpSpam: 92\.52%, DeRev2014: 100%\), while news datasets show intermediate difficulty, with GPT\-4o achieving 65\.92% zero\-shot on FakeNewsNet, likely benefiting from world knowledge\. Figure[1](https://arxiv.org/html/2607.29066#S4.F1)summarises the best achievable F1 per method per dataset, illustrating the consistent advantage of fine\-tuning over LLM\-based approaches across six of the seven datasets, with the exception of RLTD where few\-shot LLMs prevail\.

![Refer to caption](https://arxiv.org/html/2607.29066v1/x1.png)Figure 1\.Best F1 per method per dataset\.
##### Impact of Few\-shot Learning

Few\-shot prompting yields inconsistent gains\. GPT\-4o improves substantially on RLTD \(41\.59% to 84\.62%\) and DECOUR \(40\.52% to 66\.16%\) with 4\-shot Top\-K\. Conversely, LLaMA models degrade on several datasets under few\-shot prompting\. High\-variance example selection occasionally outperforms Top\-K \(e\.g\., GPT\-4o on DeRev2014, High\-Variance \(91\.92%\) outperforms Top\-K \(87\.76%\)\) but shows no consistent advantage, confirming sensitivity to example choice\. Figure[2](https://arxiv.org/html/2607.29066#S4.F2)traces GPT\-4o and GPT\-4o\-mini across all three direct\-classification shot variants and all seven datasets; annotated values show the per\-dataset maximum \(dark blue\) and minimum \(dark red\) across all six variants\.

![Refer to caption](https://arxiv.org/html/2607.29066v1/x2.png)Figure 2\.GPT\-4o & GPT\-4o\-mini: Zero\-Shot vs Few\-Shot \(Direct Classification\)
##### Chain\-of\-Thought Reasoning

CoT shows inconsistent effects: GPT\-4o zero\-shot CoT improves over zero\-shot direct on RLTD \(41\.59% to 71\.65%\), yet 4\-shot direct \(84\.62%\) outperforms 4\-shot CoT \(62\.94%\)\. CoT consistently underperforms direct classification on OpSpam\. Dataset\-specific benefits appear on DeRev2014 \(LLaMA3\-8B CoT: 58\.00% vs\. direct: 44\.29%\)\. These findings align with\(Miahet al\.,[2025](https://arxiv.org/html/2607.29066#bib.bib21)\), confirming reasoning steps do not universally benefit deception detection\.

##### Model Size and Architecture Effects

Larger models do not guarantee better performance: GPT\-4o outperforms GPT\-4o\-mini but margins are modest\. Among fine\-tuned models, DeBERTa and DistilBERT show strongest overall performance; ALBERT lags despite architectural similarity to BERT\. T5\-base underperforms on small datasets \(RLTD: 14\.48%\) but remains competitive on larger corpora\.

##### Stability Analysis \(RLTD\)

Given RLTD’s small size, we performed 10\-fold cross\-validation \(Table[4](https://arxiv.org/html/2607.29066#S4.T4)\)\. High standard deviations confirm substantial instability: BERT improves from 48\.11% to 62\.86% CV mean, while DeBERTa drops from 77\.53% to 61\.46%, indicating overfitting on single splits\. Among LLMs, GPT\-4o with Top\-K selection achieves the highest CV mean \(78\.14%, std=0\.053\), the most stable strong result across all configurations\.

##### Temperature Sensitivity \(RLTD\)

Table[5](https://arxiv.org/html/2607.29066#S4.T5)shows LLM performance across temperatures \(0\.0–1\.0\) on RLTD\. Most models exhibit low sensitivity: Gemma2\-9B few\-shot Top\-K varies only 1\.06 std across temperatures\. LLaMA3\.1\-8B shows higher variance under few\-shot High\-Variance selection \(std=4\.32\), suggesting interaction between example diversity and output stochasticity\. Phi\-3\-mini produces near\-constant output across temperatures in zero\-shot settings \(std=0\.65 across folds\), suggesting degenerate behaviour\.

## 5\.Discussion

After achieving the 100% F1 score for DeRev2014 across multiple models, we looked for existing work where similar results are reported\.Papantoniouet al\.\([2022](https://arxiv.org/html/2607.29066#bib.bib115)\)report identical results and identifies lexical leakage: words such asthrillerappear exclusively in deceptive samples, whileStephenappears only in truthful ones, arising from the use of different books per class\. This constitutes a dataset quality artefact rather than genuine generalisation, underscoring the need for carefully constructed benchmarks\.

Our results confirm strong domain dependence with limited cross\-domain transfer\. Fine\-tuned models optimised on review data cannot be deployed in legal contexts without substantial degradation, consistent with prior findings on domain\-specific deceptive cues\(Sarzynska\-Waweret al\.,[2023](https://arxiv.org/html/2607.29066#bib.bib41); Loconteet al\.,[2025](https://arxiv.org/html/2607.29066#bib.bib47)\)\. This motivates domain\-adaptive training or multi\-domain fine\-tuning pipelines\.

A clear trade\-off emerges between fine\-tuning and prompting\. Datasets with sufficient samples and stylistic regularity \(OpSpam, DECOUR\) benefit from fine\-tuning\. On small legal datasets \(RLTD: 121 samples\), few\-shot GPT\-4o \(84\.62%\) avoids the overfitting risks of fine\-tuning\. For legal practitioners with limited annotations, few\-shot LLMs offer a practical entry point; organisations with annotation capacity should invest in fine\-tuned models for cost\-effective deployment\.

Counterintuitively, explicit reasoning via CoT often degrades performance relative to direct classification\. This suggests that deception detection relies more on pattern recognition than on systematic logical inference, contrasting with CoT’s effectiveness on tasks such as mathematical reasoning and highlighting the need for task\-specific prompt engineering\.

## 6\.Conclusion

We present a survey and comparative evaluation of NLP\-based deception detection with emphasis on legal and law enforcement contexts\. Experiments across seven datasets and thirteen models under multiple prompting strategies yield several actionable findings\. We demonstrate that model performance is highly domain\-dependent necessitating domain\-specific adaptation\. Fine\-tuning outperforms prompting given sufficient data, but few\-shot GPT\-4o achieves competitive performance \(84\.62% F1\) on small legal datasets\. Chain\-of\-thought prompting shows inconsistent effects, often underperforming direct classification, challenging assumptions on universal utility of reasoning\. No single model dominates across all datasets; optimal selection depends on domain, dataset size, and linguistic complexity\. For legal practitioners, current ADD systems should serve as decision\-support tools rather than autonomous mechanisms\. Critical gaps remain in interpretability, bias mitigation, and adversarial robustness\. Future work should prioritise domain adaptation, interpretable fine\-tuning, and the construction of large\-scale legal deception corpora with rich metadata to advance trustworthy deployment\.

## References

- M\. Aspromonte, G\. Contissa, F\. Galli, and A\. Loreggia \(2025\)Beyond fact\-checking: a scalable, domain\-agnostic, and explainable system for automated fake news detection\.Proceedings of ROMCIR,pp\. 70–78\.Cited by:[§3\.1\.3](https://arxiv.org/html/2607.29066#S3.SS1.SSS3.p1.1)\.
- J\. J\. Benny \(2023\)Knowledge informed fake news detection using large language models\.Master’s Thesis,University of Windsor \(Canada\)\.Cited by:[§3\.1\.3](https://arxiv.org/html/2607.29066#S3.SS1.SSS3.p1.1)\.
- K\. Chen, Z\. Lian, H\. Sun, R\. Liu, J\. Yi, B\. Liu, and J\. Tao \(2024\)Can deception detection go deeper? dataset, evaluation, and benchmark for deception reasoning\.arXiv preprint arXiv:2402\.11432\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.10.9.1.1.1)\.
- Z\. Cui, T\. Huang, C\. Chiang, and C\. Du \(2025\)Toward verifiable misinformation detection: a multi\-tool llm agent framework\.InProceedings of the 2025 International Conference on Generative Artificial Intelligence for Business,pp\. 179–185\.Cited by:[§3\.1\.3](https://arxiv.org/html/2607.29066#S3.SS1.SSS3.p1.1)\.
- T\. Fornaciari, F\. Bianchi, M\. Poesio, D\. Hovy,et al\.\(2021\)BERTective: language models and contextual information for deception detection\.InProceedings of the 16th conference of the european chapter of the association for computational linguistics: Main volume,Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.8.7.1.1.1)\.
- T\. Fornaciari, M\. Poesio,et al\.\(2012\)DeCour: a corpus of deceptive statements in italian courts\.\.InLREC,pp\. 1585–1590\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.6.5.4.1.1),[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.7.6.4.1.1),[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.8.7.4.1.1),[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px2),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.3.2.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.4.1.1.1)\.
- T\. Fornaciari and M\. Poesio \(2012\)On the use of homogenous sets of subjects in deceptive language analysis\.InProceedings of the Workshop on Computational Approaches to Deception Detection,pp\. 39–47\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.6.5.1.1.1)\.
- T\. Fornaciari and M\. Poesio \(2013\)Automatic deception detection in italian court cases\.Artificial intelligence and law21\(3\),pp\. 303–340\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.7.6.1.1.1)\.
- T\. Fornaciari and M\. Poesio \(2014\)Identifying fake amazon reviews as learning from crowds\.InProceedings of the 14th Conference of the European Chapter of the Association for Computational Linguistics,pp\. 279–287\.Cited by:[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px3.p1.1),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.6.5.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.7.1.1.1)\.
- J\. Li, W\. Chen, Q\. Xu, N\. Shah, J\. C\. Kohler, and T\. K\. Mackey \(2020\)Detection of self\-reported experiences with corruption on twitter using unsupervised machine learning\.Social Sciences & Humanities Open2\(1\),pp\. 100060\.Cited by:[§2\.1](https://arxiv.org/html/2607.29066#S2.SS1.p1.1)\.
- R\. Loconte, C\. Battaglini, S\. Maldera, P\. Pietrini, G\. Sartori, N\. Navarin, and M\. Monaro \(2025\)Detecting deception through linguistic cues: from reality monitoring to natural language processing\.Journal of Language and Social Psychology44\(3\-4\),pp\. 523–552\.Cited by:[§5](https://arxiv.org/html/2607.29066#S5.p2.1)\.
- M\. M\. M\. Miah, A\. Anika, X\. Shi, and R\. Huang \(2025\)Hidden in plain sight: evaluation of the deception detection capabilities of llms in multimodal settings\.arXiv preprint arXiv:2506\.09424\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.5.4.1.1.1),[§3\.1\.3](https://arxiv.org/html/2607.29066#S3.SS1.SSS3.p1.1),[§4](https://arxiv.org/html/2607.29066#S4.SS0.SSS0.Px4.p1.1)\.
- T\. Nguyen, F\. Abri, A\. S\. Namin, and K\. S\. Jones \(2024\)Deception and lie detection using reduced linguistic features, deep models and large language models for transcribed data\.In2024 IEEE 48th Annual Computers, Software, and Applications Conference \(COMPSAC\),pp\. 376–381\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.4.3.1.1.1)\.
- M\. Ott, Y\. Choi, C\. Cardie, and J\. T\. Hancock \(2011\)Finding deceptive opinion spam by any stretch of the imagination\.arXiv preprint arXiv:1107\.4557\.Cited by:[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px3.p1.1.1),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.4.3.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.5.1.1.1)\.
- K\. Papantoniou, P\. Papadakos, T\. Patkos, G\. Flouris, I\. Androutsopoulos, and D\. Plexousakis \(2022\)Deception detection in text and its relation to the cultural dimension of individualism/collectivism\.Natural Language Engineering28\(5\),pp\. 545–606\.Cited by:[§5](https://arxiv.org/html/2607.29066#S5.p1.1)\.
- K\. Papantoniou, P\. Papadakos, and D\. Plexousakis \(2025\)Evaluating llms on deceptive text across cultures\.InRANLP,pp\. 884–893\.Cited by:[§3\.1\.3](https://arxiv.org/html/2607.29066#S3.SS1.SSS3.p1.1)\.
- V\. Pérez\-Rosas, M\. Abouelenien, R\. Mihalcea, and M\. Burzo \(2015\)Deception detection using real\-life trial data\.InProceedings of the 2015 ACM on international conference on multimodal interaction,pp\. 59–66\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.2.1.1.1.1),[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.2.1.4.1.1),[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.3.2.4.1.1),[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.4.3.4.1.1),[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.5.4.4.1.1),[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px1),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.2.1.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.3.1.1.1)\.
- V\. Pérez\-Rosas and R\. Mihalcea \(2014\)Cross\-cultural deception detection\.InProceedings of the 52nd Annual Meeting of the Association for Computational Linguistics \(Volume 2: Short Papers\),pp\. 440–445\.Cited by:[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px3.p1.1),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.5.4.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.6.1.1.1)\.
- L\. Quijano\-Sánchez, F\. Liberatore, J\. Camacho\-Collados, and M\. Camacho\-Collados \(2018\)Applying automatic text\-based detection of deceptive language to police reports: extracting behavioral patterns from a multi\-step classification model to understand how we lie to the police\.Knowledge\-Based Systems149,pp\. 155–168\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.9.8.1.1.1)\.
- J\. Sarzynska\-Wawer, A\. Pawlak, J\. Szymanowska, K\. Hanusz, and A\. Wawer \(2023\)Truth or lie: exploring the language of deception\.Plos one18\(2\),pp\. e0281179\.Cited by:[§5](https://arxiv.org/html/2607.29066#S5.p2.1)\.
- K\. Shu, D\. Mahudeswaran, S\. Wang, D\. Lee, and H\. Liu \(2020\)Fakenewsnet: a data repository with news content, social context, and spatiotemporal information for studying fake news on social media\.Big data8\(3\),pp\. 171–188\.Cited by:[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px3.p1.1),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.8.7.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.9.1.1.1)\.
- A\. Velutharambath and R\. Klinger \(2023\)UNIDECOR: a unified deception corpus for cross\-corpus deception detection\.arXiv preprint arXiv:2306\.02827\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.3.2.1.1.1),[§3\.1\.3](https://arxiv.org/html/2607.29066#S3.SS1.SSS3.p1.1)\.
- W\. Y\. Wang \(2017\)” Liar, liar pants on fire”: a new benchmark dataset for fake news detection\.arXiv preprint arXiv:1705\.00648\.Cited by:[§3\.1\.1](https://arxiv.org/html/2607.29066#S3.SS1.SSS1.Px3.p1.1),[§3\.1\.2](https://arxiv.org/html/2607.29066#S3.SS1.SSS2.p1.2),[Table 2](https://arxiv.org/html/2607.29066#S3.T2.7.1.7.6.1),[Table 3](https://arxiv.org/html/2607.29066#S4.T3.1.1.1.1.8.1.1.1)\.
- C\. Xiao, H\. Zhong, Z\. Guo, C\. Tu, Z\. Liu, M\. Sun, Y\. Feng, X\. Han, Z\. Hu, H\. Wang,et al\.\(2018\)Cail2018: a large\-scale legal dataset for judgment prediction\.arXiv preprint arXiv:1807\.02478\.Cited by:[Table 1](https://arxiv.org/html/2607.29066#S2.T1.1.1.10.9.4.1.1)\.
- X\. Zhang and W\. Gao \(2023\)Towards llm\-based fact verification on news claims with a hierarchical step\-by\-step prompting method\.arXiv preprint arXiv:2310\.00305\.Cited by:[§2\.1](https://arxiv.org/html/2607.29066#S2.SS1.p2.1)\.