NIMS: MagIC - Artigos em revista internacional com arbitragem científica (Peer-Review articles in international journals)
URI permanente para esta coleção:
Navegar
Entradas recentes
- Dissociative Electron Attachment Prediction of Halogenated Organic Molecules Using Machine LearningPublication . Silva, Tomas; Lobo, Victor Sousa; Pereira-da-Silva, Joao; Da Silva, Filipe Ferreira; NOVA Information Management School (NOVA IMS); Information Management Research Center (MagIC) - NOVA Information Management School; DF – Departamento de Física; CeFITec – Centro de Física e Investigação Tecnológica; ACS - American Chemical SocietyDissociative Electron Attachment (DEA) is a fundamental process in which low-energy electrons interact with molecules, causing bond dissociation and the formation of negative ions. It plays a key role in environmental science, nanotechnology, biology, and astrochemistry. However, experimental DEA studies are typically conducted in the gas phase under high-vacuum conditions, limiting the investigation of larger or less volatile molecules. To address these limitations, we developed machine learning (ML) models to predict negative ion formation in halogenated organic molecules. Classification models were designed to predict the energy range of the most intense DEA resonance, while regression models estimate its peak energy. A relational database consolidating experimental DEA data and molecular descriptors for 143 molecules served as the basis for model training and evaluation. Various ML approaches spanning different algorithm families were compared, using 120 molecules for training and 23 for testing. The ensemble Voting Classifier, combining three models from different families, achieved 94.9% accuracy in cross-validation with only one test set misclassification. The Random Forest model achieved the best regression results with a mean absolute error of 0.301 eV in cross-validation and 0.234 eV on the test set, comparable to typical experimental resolutions. These results demonstrate the feasibility of ML-based DEA prediction, establishing a foundation for computational approaches capable of expanding research to molecules that are challenging to study experimentally.
- No one is watchingPublication . Espartel, Lélis Balestrin; Rohden , Simoni F.; Araújo, Clécio Falcão; Information Management Research Center (MagIC) - NOVA Information Management School; NOVA Information Management School (NOVA IMS); ElsevierThe growing use of artificial intelligence (AI) agents in frontline service roles is reshaping frontline interactions, with important implications for consumer ethical behavior. This research examines whether AI-mediated service interactions increase unethical consumer behavior and identifies the psychological mechanism underlying this effect. Across a pilot study and three experiments, we show that consumers who interact with AI agents report higher unethical behavioral intentions than those who interact with human employees. This effect is explained by reduced perceived social judgment, defined as the extent to which consumers perceive the interacting agent as capable of morally evaluating their behavior, rather than by entitlement, anticipatory guilt, or moral disengagement. The effect is strongest when both internal (moral identity) and external (detectability) accountability are low. Importantly, cognitive priming of evaluative concern does not restore this perception in AI-mediated interactions, suggesting that reduced social judgment is anchored in the agent’s perceived evaluative capacity rather than in momentary cognitive salience. These findings show that AI alters the evaluative dynamics of retail and service encounters, increasing the risk of unethical consumer behavior, and highlight the need for firms to embed accountability mechanisms when deploying AI in customer-facing roles.
- Beyond the promptPublication . Godinho, Sofia; Oliveira, Tiago; Neves, Catarina; NOVA Information Management School (NOVA IMS); Information Management Research Center (MagIC) - NOVA Information Management School; Associação Portuguesa para o Estudo do Quaternário (APEQ)Purpose As the use of ChatGPT continues to expand, understanding its impact on users remains an emerging area of study. This research investigates how individuals form a sense of identity around ChatGPT, applying the information technology (IT) identity framework, with an added outcomes component, to assess the psychological relationship users develop with the tool. Design/methodology/approach A survey of 312 active ChatGPT users was conducted to evaluate the framework’s applicability in this context and to explore the resulting behavioral and experiential outcomes. The model was tested using partial least squares structural equation modeling. Findings The results provide partial support for the IT identity framework in the ChatGPT context. The analysis indicates that regular interaction with ChatGPT is associated with the development of a distinct IT identity. Moreover, the presence of sufficient resources and organizational or contextual support significantly moderates this relationship, shaping user behaviors in meaningful ways. These behaviors, in turn, were linked to improvements in both individual performance and perceived social well-being. The findings also suggest that some relationships established in traditional IT contexts may operate differently in generative AI environments. Originality/value By extending the IT identity framework to include specific outcome measures, this study offers a nuanced understanding of how AI-based tools like ChatGPT influence user identity and behavior based on the IT identity framework. The findings further contribute to understanding how AI-related identities form and highlight opportunities for refining IT identity theory in the context of generative AI assistants.
- Semantic Drift in Long-Form Financial Disclosures in PortuguesePublication . Correa, João Victor M.; Bravo, Jorge Miguel; Silveira, Rodrigo Lanna Franco da; Cruz Júnior, José César; Batista, Fernando; Moraes Silva, Renato; Information Management Research Center (MagIC) - NOVA Information Management School; NOVA Information Management School (NOVA IMS); ElsevierFinancial disclosures contain critical information that is not always immediately reflected in market prices. Tracking semantic change in these communications can surface narrative shifts, but it is difficult because financial documents are long, templated, and evolve gradually, which blurs meaningful drift with routine variation. Moreover, Portuguese-language financial disclosures remain underrepresented in the financial natural language processing literature. We propose a framework for measuring semantic drift in long-form Portuguese disclosures using document embeddings, chunking, and aggregation, with drift defined as consecutive cosine distance. Article-scope inference uses 3 models and 5 aggregation techniques (15 configurations per dataset) across five datasets spanning corporate and public-sector reporting. We validate drift against volatility signals using circular-shift tests, filing-date time windows, and Granger causality, complemented by descriptive event alignment and event-study diagnostics. We find that Vale exhibits a positive drift–volatility association under within-model correction, EDP shows robust drift-to-volatility Granger predictability across configurations, and SLC shows a robust filing-date window association. Conab exhibits a similar Granger pattern only when mapped to SLC stock as market proxy, making that result proxy-sensitive. Other model-dataset combinations show weaker or non-significant links, highlighting sensitivity to document type and template structure.
- Assessing terrestrial vertebrates as potential biodiversity indicators of ecosystem services in multifunctional landscapesPublication . Campos, Felipe S.; David, João; Lourenço-De-Moraes, Ricardo; Fuzessy, Lisieux; Síllero, Neftalí; Cabral, Pedro; Retana, Javier; Information Management Research Center (MagIC) - NOVA Information Management School; Wiley-BlackwellCombining ecosystem services and biodiversity conservation is critical for evaluating the benefits of nature to society. However, research gaps have led to a conservation dilemma in determining which biodiversity groups should be selected in ecosystem service assessments. To contribute to the ongoing debate on the role of biodiversity in supporting ecosystem services, we assess the performance of terrestrial vertebrates as biodiversity indicators of ecosystem services based on their spatial alignment across landscapes. Using mainland Portugal as a case study, we model the spatial supply of eight ecosystem services with the Integrated Valuation of Ecosystem Services and Trade-offs (InVEST) toolkit and related land cover-based spatial models. Using a transferable spatial framework, we apply the indicator value index to map how biodiversity and ecosystem service targets co-occur in multifunctional landscapes where natural and semi-natural ecosystems coexist with long-term human land uses. Based on species distribution data for amphibians, reptiles, birds and mammals, we identify reptiles as the vertebrate group most strongly associated with ecosystem services. Our results show that reptiles are spatially associated with all eight ecosystem services assessed and account for around 40% of the total indicator value scores. In contrast, the other vertebrate groups are each associated with four services, and habitat quality is the only ecosystem service common to all groups. These findings suggest that terrestrial vertebrates, especially reptiles, may serve as potential biodiversity indicators of multiple ecosystem service supply patterns. Under the Post-2020 Global Biodiversity Framework, our findings introduce a GIS-based approach for a better understanding of relationships between biodiversity and ecosystem services, including their representativeness in protected areas, according to the European Natura 2000 network. The study also contributes to an ongoing national effort to map the supply of multiple ecosystem services based on sets of possible indicator values derived from the landscape.
- Leveraging large language models for agentic process analytics assistantsPublication . Reis, Diogo; Caldeira, João; Jesus, Marta; NOVA Information Management School (NOVA IMS); Elsevier Science B.V., Amsterdam.To leverage Large Language Models for enhanced data-driven decision-making in Process Mining, this study introduces a framework that establishes an extended process analytics question taxonomy, including case, activity, resource, temporal, discovery and variant, and what-if and predictive dimensions. The proposed framework evaluates state-of-the-art Large Language Models on their ability to generate executable Python-based visual analytics, bridging the gap between raw Process Mining insights and non-technical users. Performance is assessed for recent models, such as GPT-4.1 and Claude 3.7, using an LLM-as-Judge approach and a specific Visualisation Error Rate. The framework works both as a benchmarking instrument and as a conceptual and practical basis to guide the development of deployable agentic assistants within AI-Augmented Business Process Management Systems. Results indicate that GPT-4.1 achieves on a bounded 1-10 evaluation scale (higher is better), an LLM-as-Judge score of μ = 8.18, σ = 2.3 and a Visualisation Error Rate of 25%. A key finding reveals a significant performance disparity: while models have an increased performance with widely adopted libraries, they face substantial challenges with domain-specific tools like PM4Py. These insights provide the necessary evidence to guide model refinement to fully support automated visual insight generation in Process Mining contexts, including the ones backed by other tools and languages, such as bupaR with R, and ProM with Java.
- Unification of Closed-Open Industrial Detection ScenariosPublication . Zhang, Zekai; Zhang, Jinglin; Chen, Qinghui; Li, Gang; Chen, Da; Jing, Shuainan; Wang, He; Li, Dagang; Liu, Cong; Bai, Cong; Chen, Shengyong; NOVA Information Management School (NOVA IMS); Information Management Research Center (MagIC) - NOVA Information Management School; IEEE Computer SocietyLarge-scale Visual-Language Models (LVLMs) have achieved remarkable success in natural visual tasks, yet their application to industrial defect detection remains challenging due to two fundamental limitations: (i) the scarcity of large-scale industrial datasets that cover diverse defect categories across multiple domains, and (ii) the reliance on manual prompts (points, boxes, masks) that introduce subjective noise and lack text-visual interaction for fine-grained understanding. To address these challenges, we introduce a Large-Scale Multi-Modal Industrial Open-Closed benchmark (MMIOC-1M) containing over one million samples across 14 super-categories, 29 industrial scenes, and 351 defect subcategories. To our knowledge, MMIOC-1M is the first unified largest benchmark supporting both open-vocabulary and closed-set industrial detection, providing valuable pre-training data for LVLMs in industrial scenarios. Furthermore, we propose a Refined Text-Visual Prompt Network (RTVPNet) that incorporates three key innovations: (1) an expert-assisted domain projection mechanism that enables rapid adaptation of general vision models to industrial domains, (2) an energy-based sparse sampling strategy that automatically generates refined visual prompts without manual intervention, and (3) a bidirectional text-visual interaction module that enhances cross-modal semantic alignment and understanding. Extensive experiments demonstrate that RTVPNet achieves state-of-the-art performance on MMIOC-1M, LVIS, and COCO benchmarks while maintaining computational efficiency. The dataset and code are available at https://github.com/hellozzk/MMIO.
- Entering B2B brands' living roomsPublication . Österle, Benjamin; Sarasvuo, Sonja; Kuhn, Marc; Henseler, Jörg; Information Management Research Center (MagIC) - NOVA Information Management School; Elsevier Science B.V., Amsterdam.Brand worlds are powerful tools for branding and for creating extraordinary customer experiences in B2C markets, and they are increasingly applied in B2B contexts as well. However, the specific characteristics of industrial marketing raise the question of whether brand worlds function similarly in this setting. This study examines how visiting a brand world is related to brand experience and brand equity in industrial marketing. Drawing on data from 218 business visitors, we employed a pretest-posttest quasi-experimental design combined with structural equation modeling. The findings reveal that brand world visits are associated with higher levels of brand experience and brand equity through the multidimensional concept of brand world experience, a higher-order formative construct. Brand world experience mediates the link between pre- and post-visit brand experience, but is not associated with pre-visit brand equity. Post-visit brand equity is related both directly to the brand world experience and indirectly to post-visit brand experience. These findings demonstrate that, when embedded in experiential marketing strategies, brand worlds function as instruments for enhancing brand experience and brand equity in industrial marketing, thereby underscoring their role as the metaphorical “living room of the brand.”
- Asymmetric volatility transmission in cryptocurrency marketsPublication . Santos, Mariana; Iorio, Carmela; Damásio, Bruno; Information Management Research Center (MagIC) - NOVA Information Management School; NOVA Information Management School (NOVA IMS); Elsevier Science B.V., Amsterdam.We examine how risk travels between large-cap and small-cap cryptocurrencies and how major news shocks amplify these linkages. Using daily data for nine large-cap/small-cap pairs from September 2018–March 2023, we combine a multivariate volatility model with an event study of eleven major episodes, including the Terra-Luna collapse. Three results emerge. First, volatility transmission runs mainly from large-cap coins to smaller ones: 7 of 9 high-to-low cross-shock coefficients are statistically significant, versus 3 in the opposite direction. Second, negative events trigger larger abnormal returns and more frequent significant post-event effects than positive events, especially among small-cap coins. Third, conditional correlations rise during stress, pointing to stronger market integration and weaker diversification exactly when it matters most. The implied hedge ratios show that short positions in large-cap coins can partly protect small-cap exposure. As cryptocurrency markets mature, monitoring these transmission channels will remain important for portfolio design and for the regulation of systemic digital-asset risk.
- Online Multi-Task Business Process Prediction Using Dynamic RepresentationPublication . Zhang, Xiwei; Fang, Xianwen; Bao, Wei; Liu, Cong; NOVA Information Management School (NOVA IMS); Information Management Research Center (MagIC) - NOVA Information Management School; Institute of Electrical and Electronics Engineers (IEEE)Existing Predictive Process Monitoring (PPM) methods typically rely on static offline methods, limiting their ability to adapt to dynamic process evolution driven by emerging activities and shifting behavioral patterns. This constraint is particularly critical for multi-task prediction, such as the simultaneous forecasting of the next activity and the remaining process time. To address this challenge, we propose an online multi-task prediction framework based on dynamic graph representations. The framework enables a Graph Neural Network (GNN) to incrementally learn newly emerging activities by leveraging dynamic graph snapshots and an architecture expansion strategy. For efficient online adaptation, the framework incorporates two update strategies, a standard periodic update and a drift-aware adaptive update triggered by the Maximum Mean Discrepancy (MMD2) between subgraph embeddings. Both strategies are integrated with a Prioritized Experience Replay (PER) mechanism, augmented with a rarity-aware bonus, to ensure rapid and robust model adjustments in non-stationary environments. Comprehensive experiments on multiple real-world event logs demonstrate that our framework, when combined with different GNN backbones such as GCN, GAT, and GIN, significantly outperforms state-of-the-art baselines in both next-activity and remaining time prediction. Notably, under concept drift, the proposed drift-aware strategy exhibits strong adaptability, highlighting the framework’s effectiveness and potential for addressing complex online process prediction challenges.