Abstract
Hierarchical document classification is essential for structuring large-scale textual corpora in domains such as digital libraries and academic repositories. While recent advances in large language models (LLMs) have opened new possibilities for text classification, their applicability to hierarchical settings under real-world constraints remains underexplored. This study investigates both generative and discriminative transformer-based models, evaluating their effectiveness across multiple inference strategies: zero-shot baseline, local fine-tuning, and a global approach using category-specific models. Experiments on two real-world hierarchical datasets provide a comprehensive comparison of classification accuracy, F1-macro scores, and inference times. The results highlight that, although generative LLMs can deliver competitive (yet variable) performance at higher levels of the hierarchy, their high inference costs hinder their use in time-sensitive applications. In contrast, fine-tuned discriminative models—particularly BERT-based architectures—consistently offer a more favorable trade-off between performance and efficiency. © 2025 Elsevier B.V., All rights reserved.

Abstract
Topological photonics, leveraging concepts from condensed matter physics, offers transformative potential in the design of robust optical systems. This study investigates the integration of topologically protected edge states into plasmonic nanostructures for enhanced optical sensing. We propose a toy model comprising two chains of metallic filaments forming a one-dimensional plasmonic crystal with diatomic-like unit cells, positioned on a waveguide. The system exhibits edge states localized at the boundaries and a central defect, supported by the Su-Schrieffer-Heeger (SSH) model. These edge states, characterized by significant electric field enhancement and topological robustness, are shown to overcome key limitations in traditional plasmonic sensors, including sensitivity to noise and fabrication inconsistencies. Through coupled mode theory, we demonstrate the potential for strong coupling between plasmonic and guided optical modes, offering pathways for improved interferometric sensing schemes. This work highlights the applicability of topological photonics in advancing optical sensors.

Abstract
In recent years, the development of Decision Support Systems (DSS) has played an instrumental role in the advancement of offshore renewable energy projects, particularly within the blue energy sector. Notwithstanding the technological advancements that have been made, the acceleration of such projects continues to be impeded by significant obstacles related to stakeholder engagement, feasibility assessment, and policy compliance. The objective of this study is to propose a design for a DSS for accelerating the construction of blue offshore energy platforms. This is to address the aforementioned challenges by integrating insights from stakeholder feedback and innovation trends. A participatory action study was conducted through a workshop with a diverse group of experts (n=20), including policymakers, practitioners, researchers, and public entities involved in offshore energy projects. The evaluation facilitated the determination of the DSS's efficacy in addressing user requirements and the identification of areas for enhancement. This study proposes a model for integrating stakeholder insights into technological solutions for offshore energy installations, thus offers significant contributions to the domain of sustainable blue energy development.

Abstract
We study the dynamics as well as the interaction of stable dissipative solitons (DSs) of the cubic complex Ginzburg-Landau equation which are stabilized only by nonlinear gradient (NLG) terms. First we review stationary, periodic, quasi-periodic, and chaotic solutions. Then we investigate sudden transitions to chaotic from periodic and vice versa as a function of one parameter, as well as different outcomes, for fixed parameters, when varying the initial condition. In addition, we present a quasi-analytic approach to evaluate the separation of nearby trajectories for the case of stationary DSs as well as for periodic DSs, both stabilized by nonlinear gradient terms. In a separate section collisions between different types of DSs are reviewed. First we present a concise review of collisions of DSs without NLG terms and then the results of collisions between stationary DSs stabilized by NLG terms are summarized focusing on the influence of the nonlinear gradient term associated with the Raman effect. We point out that both, meandering oscillatory bound states as well as bound states with large amplitude oscillations appear to be specific for coupled cubic complex Ginzburg-Landau equations with a stabilizing cubic nonlinear gradient term.

Abstract
This study introduces the concept of DEEPEIA, a novel deep learning (DL) platform designed to recommend the optimal export market, and its ideal foreign champion, for any product or service offered by a small and medium-sized enterprise (SME). Drawing on expertise in SME internationalization and leveraging recent advances in generative artificial intelligence (AI), this research addresses key challenges faced by SMEs in global expansion. A systematic review of existing platforms was conducted to identify current gaps and inform the conceptualization of an advanced generative DL recommender system. The Discussion section proposes the conceptual framework for such a decision optimizer within the context of contemporary technological advancements and actionable insights. The conclusion outlines future research directions, practical implementation strategies, and expected obstacles. By mapping the current landscape and presenting an original forecasting tool, this work advances the field of AI-enabled SME internationalization while still acknowledging that more empirical validation remains a necessary next step.

Abstract
Traditional machine learning techniques assume that data is drawn from a stationary source. This assumption is challenged in contexts with data streams for presenting constant and potentially infinite sequences whose distribution is prone to change over time. Based on these settings, detecting changes (a.k.a. concept drifts) is necessary to keep learning models up-to-date. Although state-of-the-art detection methods were designed to monitor the loss of predictive models, such monitoring falls short in many real-world scenarios where the true labels are not readily available. Therefore, there is increasing attention to unsupervised concept drift detection methods as approached in this paper. In this work, we present an unsupervised and interpretable method based on Radial Basis Function Networks (RBFN) and Markov Chains (MC), referred to as RMIDDM (Radial Markov Interpretable Drift Detection Method). In our method, RBF performs, in the intermediate layer, an activation process that implicitly produces groups of observations collected over time. Simultaneously, MC models the transitions between groups to support the detection of concept drifts, which happens when the active group changes and its probability exceeds a given threshold. A set of experiments with synthetic datasets and comparisons with state-of-the-art algorithms demonstrated that the proposed method can detect drifts at runtime in an efficient, interpretable, and independent way of labels, presenting competitive results and behavior. Additionally, to show its applicability in a real-world scenario, we analyzed new COVID-19 cases, deaths, and vaccinations to identify new waves as concept drifts and generate Markov models that allow understanding of their interaction.