arXiv subjects

Benjamin Icard

Publications and source records attributed to Benjamin Icard.

14 recordsLinked to original sources

arXiv2026

Propaganda Forensics: Recovering the Generation Pipeline of an AI-Driven Influence Campaign

We present a forensic analysis of the generation pipeline behind a recent AI-driven influence campaign. We introduce PROPAGIA, a corpus of 2,646 propagandist French articles from the Storm-1516/CopyCop campaign disclosed by VIGINUM and INSIKT GROUP in 2025. For comparison, we rely on SIPA, a corpus of human-written French mainstream press from the same period. Using topic modeling, vagueness and sentiment analysis, we first isolate persuasion techniques characteristic of propaganda, with PROPAGIA far exceeding SIPA in vagueness, subjectivity and negativity, and citing fewer sources. We then find prompt instruction leaks on 50 of the 84 PROPAGIA websites, including a verbatim ten-point editorial specification accounting for several of these differences, together with high cross-article redundancy. Finally, we show that rewriting-based detection supports INSIKT GROUP's attribution to the Llama 3 family, but also suggests the involvement of Mistral-family models.

cs.AI↗

arXiv2026

Measuring Embedding Sensitivity to Authorial Style in French: Comparing Literary Texts with Language Model Rewritings

Large language models (LLMs) can convincingly imitate human writing styles, yet it remains unclear how much stylistic information is encoded in embeddings from any language model and retained after LLM rewriting. We investigate these questions in French, using a controlled literary dataset to quantify the effect of stylistic variation via changes in embedding dispersion. We observe that embeddings reliably capture authorial stylistic features and that these signals persist after rewriting, while also exhibiting LLM-specific patterns. These analytical results offer promising directions for authorship imitation detection in the era of language models.

cs.CL↗

arXiv2026

Reliable News or Propagandist News? A Neurosymbolic Model Using Genre, Topic, and Persuasion Techniques to Improve Robustness in Classification

Among news disorders, propagandist news are particularly insidious, because they tend to mix oriented messages with factual reports intended to look like reliable news. To detect propaganda, extant approaches based on Language Models such as BERT are promising but often overfit their training datasets, due to biases in data collection. To enhance classification robustness and improve generalization to new sources, we propose a neurosymbolic approach combining non-contextual text embeddings (fastText) with symbolic conceptual features such as genre, topic, and persuasion techniques. Results show improvements over equivalent text-only methods, and ablation studies as well as explainability analyses confirm the benefits of the added features. Keywords: Information disorder, Fake news, Propaganda, Classification, Topic modeling, Hybrid method, Neurosymbolic model, Ablation, Robustness

cs.CL↗

arXiv2026

From Noise to Signal: When Outliers Seed New Topics

Outliers in dynamic topic modeling are typically treated as noise, yet we show that some can serve as early signals of emerging topics. We introduce a temporal taxonomy of news-document trajectories that defines how documents relate to topic formation over time. It distinguishes anticipatory outliers, which precede the topics they later join, from documents that either reinforce existing topics or remain isolated. By capturing these trajectories, the taxonomy links weak-signal detection with temporal topic modeling and clarifies how individual articles anticipate, initiate, or drift within evolving clusters. We implement it in a cumulative clustering setting using document embeddings from eleven state-of-the-art language models and evaluate it retrospectively on HydroNewsFr, a French news corpus on the hydrogen economy. Inter-model agreement reveals a small, high-consensus subset of anticipatory outliers, increasing confidence in these labels. Qualitative case studies further illustrate these trajectories through concrete topic developments.

cs.CL↗

arXiv2025

From Outliers to Topics in Language Models: Anticipating Trends in News Corpora

This paper examines how outliers, often dismissed as noise in topic modeling, can act as weak signals of emerging topics in dynamic news corpora. Using vector embeddings from state-of-the-art language models and a cumulative clustering approach, we track their evolution over time in French and English news datasets focused on corporate social responsibility and climate change. The results reveal a consistent pattern: outliers tend to evolve into coherent topics over time across both models and languages.

cs.CL↗

arXiv2025

Embedding Style Beyond Topics: Analyzing Dispersion Effects Across Different Language Models

This paper analyzes how writing style affects the dispersion of embedding vectors across multiple, state-of-the-art language models. While early transformer models primarily aligned with topic modeling, this study examines the role of writing style in shaping embedding spaces. Using a literary corpus that alternates between topics and styles, we compare the sensitivity of language models across French and English. By analyzing the particular impact of style on embedding dispersion, we aim to better understand how language models process stylistic information, contributing to their overall interpretability.

cs.CL↗

arXiv2024

HYBRINFOX at CheckThat! 2024 -- Task 2: Enriching BERT Models with the Expert System VAGO for Subjectivity Detection

This paper presents the HYBRINFOX method used to solve Task 2 of Subjectivity detection of the CLEF 2024 CheckThat! competition. The specificity of the method is to use a hybrid system, combining a RoBERTa model, fine-tuned for subjectivity detection, a frozen sentence-BERT (sBERT) model to capture semantics, and several scores calculated by the English version of the expert system VAGO, developed independently of this task to measure vagueness and subjectivity in texts based on the lexicon. In English, the HYBRINFOX method ranked 1st with a macro F1 score of 0.7442 on the evaluation data. For the other languages, the method used a translation step into English, producing more mixed results (ranking 1st in Multilingual and 2nd in Italian over the baseline, but under the baseline in Bulgarian, German, and Arabic). We explain the principles of our hybrid approach, and outline ways in which the method could be improved for other languages besides English.

cs.CL↗

arXiv2024

HYBRINFOX at CheckThat! 2024 -- Task 1: Enhancing Language Models with Structured Information for Check-Worthiness Estimation

This paper summarizes the experiments and results of the HYBRINFOX team for the CheckThat! 2024 - Task 1 competition. We propose an approach enriching Language Models such as RoBERTa with embeddings produced by triples (subject ; predicate ; object) extracted from the text sentences. Our analysis of the developmental data shows that this method improves the performance of Language Models alone. On the evaluation data, its best performance was in English, where it achieved an F1 score of 71.1 and ranked 12th out of 27 candidates. On the other languages (Dutch and Arabic), it obtained more mixed results. Future research tracks are identified toward adapting this processing pipeline to more recent Large Language Models.

cs.CL↗

arXiv2024

A Dynamic Logic for Information Evaluation in Intelligence

In the field of human intelligence, officers use an alphanumeric scale, known as the Admiralty System, to rate the credibility of messages and the reliability of their sources (NATO AJP-2.1, 2016). During this evaluation, they are expected to estimate the credibility and reliability dimensions independently of each other (NATO STANAG, 2003). However, empirical results show that officers perceive these dimensions as strongly correlated (Baker et al., 1968). More precisely, they consider credibility as playing the leading role over reliability, the importance of which is only secondary (Samet, 1975). In this paper, we present a formal evaluative procedure, called L(intel), in line with these findings. We adapt dynamic belief revision to make credibility the main dimension of evaluation and introduce dynamic operators to update credibility ratings with the source's reliability. In addition to being empirically sound, we show that L(intel) provides an effective procedure to classify intelligence messages along the descriptive taxonomy presented in Icard (2023).

cs.LO↗

arXiv2024

A Multi-Label Dataset of French Fake News: Human and Machine Insights

We present a corpus of 100 documents, OBSINFOX, selected from 17 sources of French press considered unreliable by expert agencies, annotated using 11 labels by 8 annotators. By collecting more labels than usual, by more annotators than is typically done, we can identify features that humans consider as characteristic of fake news, and compare them to the predictions of automated classifiers. We present a topic and genre analysis using Gate Cloud, indicative of the prevalence of satire-like text in the corpus. We then use the subjectivity analyzer VAGO, and a neural version of it, to clarify the link between ascriptions of the label Subjective and ascriptions of the label Fake News. The annotated dataset is available online at the following url: https://github.com/obs-info/obsinfox Keywords: Fake News, Multi-Labels, Subjectivity, Vagueness, Detail, Opinion, Exaggeration, French Press

cs.CL↗

arXiv2024

Exposing propaganda: an analysis of stylistic cues comparing human annotations and machine classification

This paper investigates the language of propaganda and its stylistic features. It presents the PPN dataset, standing for Propagandist Pseudo-News, a multisource, multilingual, multimodal dataset composed of news articles extracted from websites identified as propaganda sources by expert agencies. A limited sample from this set was randomly mixed with papers from the regular French press, and their URL masked, to conduct an annotation-experiment by humans, using 11 distinct labels. The results show that human annotators were able to reliably discriminate between the two types of press across each of the labels. We propose different NLP techniques to identify the cues used by the annotators, and to compare them with machine classification. They include the analyzer VAGO to measure discourse vagueness and subjectivity, a TF-IDF to serve as a baseline, and four different classifiers: two RoBERTa-based models, CATS using syntax, and one XGBoost combining syntactic and semantic features.

cs.CL↗

arXiv2024

Beyond the Spell: A Dynamic Logic Analysis of Misdirection

Misdirection can be defined as the intentional action of causing some misrepresentation in an agent, or in a group of agents. Such misrepresentations may result from verbal actions, as in linguistic deception, or from visual actions, as in visual misdirection. Examples of visual misdirection abound (e.g. in nature, in the military), with magic tricks providing a vivid illustration. So far, various types of verbal misdirection have been investigated from a formal perspective (e.g. lying, bluffing) but little attention has been paid to the particular case of visual misdirection. In this paper, we introduce a dynamic epistemic logic to represent not only verbal misdirection on agents' beliefs but also visual misdirection on agents' observations. We illustrate the dynamics of the logic by modelling a classic magic trick known as the French Drop. We also provide a sound and complete axiom system for the logic, and discuss the expressivity and scope of the setting.

cs.LO↗

arXiv2023

Measuring vagueness and subjectivity in texts: from symbolic to neural VAGO

We present a hybrid approach to the automated measurement of vagueness and subjectivity in texts. We first introduce the expert system VAGO, we illustrate it on a small benchmark of fact vs. opinion sentences, and then test it on the larger French press corpus FreSaDa to confirm the higher prevalence of subjective markers in satirical vs. regular texts. We then build a neural clone of VAGO, based on a BERT-like architecture, trained on the symbolic VAGO scores obtained on FreSaDa. Using explainability tools (LIME), we show the interest of this neural version for the enrichment of the lexicons of the symbolic version, and for the production of versions in other languages.

cs.CL↗

arXiv2021

Combining Vagueness Detection with Deep Learning to Identify Fake News

In this paper, we combine two independent detection methods for identifying fake news: the algorithm VAGO uses semantic rules combined with NLP techniques to measure vagueness and subjectivity in texts, while the classifier FAKE-CLF relies on Convolutional Neural Network classification and supervised deep learning to classify texts as biased or legitimate. We compare the results of the two methods on four corpora. We find a positive correlation between the vagueness and subjectivity measures obtained by VAGO, and the classification of text as biased by FAKE-CLF. The comparison yields mutual benefits: VAGO helps explain the results of FAKE-CLF. Conversely FAKE-CLF helps us corroborate and expand VAGO's database. The use of two complementary techniques (rule-based vs data-driven) proves a fruitful approach for the challenging problem of identifying fake news.

cs.CL↗