NDNicolas Dugué

Nicolas Dugué

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Nicolas Dugué has a total of 38 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Human-AI Co-design of Profile-Aware Feedback for Metacognitive Support

    by Abdelkader Ouared, Nicolas Dugué, Claudine Piau-Toffolon, Madeth May

  • Analyse du capitalisme social sur Twitter

    by Nicolas Dugué

    Le sociologue Bourdieu définit le capital social comme : "L’ensemble des ressources actuelles ou potentielles qui sont liées à la possession d’un réseau durable de relations". Sur Twitter, les abonnements, mentions et retweets créent un réseau de relations pour chaque utilisateur dont les ressources sont l’obtention d’informations pertinentes, la possibilité d’être lu, d’assouvir un besoin narcissique, de diffuser efficacement des messages.Certains utilisateurs Twitter -appelés capitalistes sociaux - cherchent à maximiser leur nombre d’abonnements pour maximiser leur capital social. Nous introduisons leurs techniques, basées sur l’échange d’abonnements et l’utilisation de hashtags dédiés. Afin de mieux les étudier, nous détaillons tout d’abord une méthode pour détecter à l’échelle du réseau ces utilisateurs en se basant sur leurs abonnements et abonnés. Puis, nous montrons avec un compte Twitter automatisé que ces techniques permettent de gagner efficacement des abonnés et de se faire beaucoup retweeter. Nous établissons ensuite que ces dernières permettent également aux capitalistes sociaux d’occuper des positions qui leur accordent une bonne visibilité dans le réseau. De plus, ces méthodes rendent ces utilisateurs influents aux yeux des principaux outils de mesure. Nous mettons en place une méthode de classification supervisée pour détecter avec précision ces utilisateurs et ainsi produire un nouveau score d’influence.

  • Automating transparent learner profiling through explainable AI

    by Abdelkader Ouared, Madeth May, Claudine Piau-Toffolon, Nicolas Dugué


  • Open Access

    Aligning Educational Stakeholder Perceptions of Learner Profiling with Explainable AI

    by Abdelkader Ouared, Madeth May, Claudine Piau-Toffolon, Nicolas Dugué

    The education community continuously develops learner profile (LP) models to support decision-making in learning analytics (LA). However, a gap persists in aligning abstract stakeholder requirements with complex machine learning (ML) patterns. Educational stakeholders such as decision-makers, educators, and pedagogical engineers perceive and categorize LPs (e.g., learners in difficulty, active/inactive learners, those in progress, and success-oriented vs. at-risk learners) through mental models. These latter reflect real-world perceptions and pedagogical practices grounded in common educational concepts. To bridge this gap, data scientists must ensure technical ML insights align with stakeholder needs by selecting relevant features, addressing explainability, mitigating biases, and validating patterns against domain assumptions. For example, a learner generating extensive log data through repeated solution attempts may appear engaged from a human perspective but exhibit disengagement based on unexpected ML discovery patterns, highlighting biases in data interpretation or human perception. We propose Req2XAI (From Requirements to Explainable Machine Learning Models), a framework that establishes a bidirectional mapping between stakeholder requirements for LP analysis and ML-driven learner profiles. Req2XAI externalizes stakeholders’ mental model about LP via a conceptual model into requirements and goals and formalizes an end-to-end workflow, from stakeholder objectives to explainable ML models, ensuring transparency at each stage. A proof-of-concept prototype is implemented through a use case, considering the requirements of the Steering Committee of the écri+ project. This work introduces open research challenges associated with the Req2XAI framework, which merit further exploration.


  • Open Access

    From communities to interpretable network and word embedding: an unified approach

    by Thibault Prouteau, Nicolas Dugué, Simon Guillot

    Modelling information from complex systems such as humans social interaction or words co-occurrences in our languages can help to understand how these systems are organized and function. Such systems can be modelled by networks, and network theory provides a useful set of methods to analyze them. Among these methods, graph embedding is a powerful tool to summarize the interactions and topology of a network in a vectorized feature space. When used in input of machine learning algorithms, embedding vectors help with common graph problems such as link prediction, graph matching, etc. Word embedding has the goal of representing the sense of words, extracting it from large text corpora. Despite differences in the structure of information in input of embedding algorithms, many graph embedding approaches are adapted and inspired from methods in NLP. Limits of these methods are observed in both domains. Most of these methods require long and resource greedy training. Another downside to most methods is that they are black-box, from which understanding how the information is structured is rather complex. Interpretability of a model allows understanding how the vector space is structured without the need for external information, and thus can be audited more easily. With both these limitations in mind, we propose a novel framework to efficiently embed network vertices in an interpretable vector space. Our Lower Dimension Bipartite Framework (LDBGF) leverages the bipartite projection of a network using cliques to reduce dimensionality. Along with LDBGF, we introduce two implementations of this framework that rely on communities instead of cliques: SINr-NR and SINr-MF. We show that SINr-MF can perform well on classical graphs and SINr-NR can produce high-quality graph and word embeddings that are interpretable and stable across runs.

  • From communities to interpretable network and word embedding: an unified approach

    by Thibault Prouteau, Nicolas Dugué, Simon Guillot

    Modelling information from complex systems such as humans social interaction or words co-occurrences in our languages can help to understand how these systems are organized and function. Such systems can be modelled by networks, and network theory provides a useful set of methods to analyze them. Among these methods, graph embedding is a powerful tool to summarize the interactions and topology of a network in a vectorized feature space. When used in input of machine learning algorithms, embedding vectors help with common graph problems such as link prediction, graph matching, etc In Natural Language Processing (NLP), such a vectorization process is also employed. Word embedding has the goal of representing the sense of words, extracting it from large text corpora. Despite differences in the structure of information in input of embedding algorithms, many graph embedding approaches are adapted and inspired from methods in NLP. Limits of these methods are observed in both domains. Most of these methods require long and resource greedy training. Another downside to most methods is that they are black-box, from which understanding how the information is structured is rather complex. Interpretability of a model allows understanding how the vector space is structured without the need for external information, and thus can be audited more easily. With both these limitations in mind, we propose a novel framework to efficiently embed network vertices in an interpretable vector space. Our Lower Dimension Bipartite Framework (LDBGF) leverages the bipartite projection of a network using cliques to reduce dimensionality. Along with LDBGF, we introduce two implementations of this framework that rely on communities instead of cliques: SINr-NR and SINr-MF. We show that SINr-MF can perform well on classical graphs and SINr-NR can produce high-quality graph and word embeddings that are interpretable and stable across runs.


  • Open Access

    Speeding up the Training of Neural Networks with the One-Step Procedure

    by Wajd Meskini, Alexandre Brouste, Nicolas Dugué

    In the last decade, research and corporate have shown a dramatically growing interest in the field of machine learning, mostly due to the performances of deep neural networks. These increasingly complex architectures solved a wide range of problems. However, training these sophisticated architectures require many computation on advanced hardware. With this paper, we introduce a new approach based on the One-Step procedure that may fasten their training. In this procedure, an initial guess estimator is computed on a subsample that is then improved with only one step of the Newton gradient descent on the whole dataset. To show the efficiency of this framework, we consider regression and classification tasks using simulated and real datasets. We consider classic architectures, namely multi-layer perceptrons and show, on our examples, that the One-Step procedure is often halving the computation time to train the neural networks while preserving the performances.


  • Filtering Communities in Word Co-Occurrence Networks to Foster the Emergence of Meaning

    by Anna Béranger, Nicolas Dugué, Simon Guillot, Thibault Prouteau


  • Open Access

    Detecting Real-World Influence Through Twitter

    by Jean-Val{È}Re Cossu, Nicolas Dugu{É, Vincent Labatut

    In this paper, we investigate the issue of detecting the real-life influence of people based on their Twitter account. We propose an overview of common Twitter features used to characterize such accounts and their activity, and show that these are inefficient in this context. In particular, retweets and followers numbers, and Klout score are not relevant to our analysis. We thus propose several Machine Learning approaches based on Natural Language Processing and Social Network Analysis to label Twitter users as Influencers or not. We also rank them according to a predicted influence level. Our proposals are evaluated over the CLEF RepLab 2014 dataset, and outmatch state-of-the-art ranking methods.