Institution profile

Universitat Politècnica de València

Academic institutioneurope · es
Official website
Research library235linked papers
Opportunities0open roles
Selected work

Representative Papers

Economic feasibility of virtual operators in 5G via network slicing

Aug 01, 2020Future generations computer systems

This study investigates incentive-compatible business models between infrastructure-holding network operators and infrastructure-less virtual operators in the context of 5G network slicing. It proposes two distinct models—strategic and monopolistic—and, for the first time, integrates a system performance model of network slicing with an economic game-theoretic framework. Resource allocation is characterized using discriminatory processor-sharing queues, and the interplay between user subscription behavior and operator pricing is analyzed through non-cooperative game theory. The findings demonstrate that the strategic model significantly enhances user subscription rates, while both models provide sufficient economic incentives for network operators to open their infrastructure to virtual operators, thereby achieving a mutually beneficial outcome for all stakeholders.

15 citationsRead paper

Generation of Standardized E-Learning Content from Digital Medical Collections

May 18, 2019Journal of medical systems

To address the heterogeneity and fragmentation of medical educational resources—which impedes seamless integration with Learning Management Systems (LMS)—this paper proposes a medical knowledge-driven multimodal content structuring and mapping framework. It enables, for the first time, fully automated conversion of heterogeneous digital medical resources (e.g., medical images, scholarly literature, clinical records) into SCORM/AICC-compliant e-learning packages. The method integrates medical ontology modeling, rule-guided template instantiation, NLP-based entity-relation extraction, and XSLT-based packaging to ensure semantic alignment and automatic generation of pedagogically sound instructional logic. Evaluated across three medical schools, the framework achieves a 12× improvement in course package generation efficiency, 100% LMS compatibility, and a 76% reduction in post-generation instructor editing time. This work establishes a reusable, high-fidelity, and standards-compliant automation paradigm for digital medical education content production.

13 citationsRead paper

Complexity of adaptive testing in scenarios defined extensionally

Oct 22, 2022Frontiers of Computer Science

This study investigates the construction of minimal adaptive test strategies that are guaranteed to verify the correctness of an implementation under test (IUT) when its behavior is explicitly defined in an enumerated manner. Focusing on four problem variants arising from the interplay of nondeterminism and multiple definitions, the work employs formal modeling and reduction techniques to systematically analyze their computational complexity. It provides the first complete characterization of the complexity landscape for adaptive testing strategies within this extended model, establishing that several variants are PSPACE-complete or Log-APX-hard. These results lay a rigorous computational complexity foundation for the theory of adaptive testing.

7 citationsRead paper

Generation of reusable learning objects from digital medical collections: An analysis based on the MASMDOA framework

Jan 01, 2021Health Informatics Journal

This study addresses the low reusability and poor adaptability of learning resources in medical education. We propose Clavy, a tool built upon the MASMDOA framework that automatically extracts, semantically restructures, and generates reusable, scenario-aware, and role-customized multimedia learning objects from heterogeneous digital medical knowledge repositories. Clavy dynamically constructs content structures based on user roles and learning objectives, and exports resources compliant with international e-learning standards (e.g., SCORM/AICC). Notably, this work introduces the first systematic application of the MASMDOA evaluation model to quantitatively assess the quality of generated medical learning objects, enabling end-to-end automated transformation from raw knowledge to standardized instructional resources. Experimental validation demonstrates Clavy’s strong compatibility with mainstream LMS platforms (e.g., Moodle, Canvas), significant gains in generation efficiency, and a 62% improvement in cross-platform resource reuse rate—establishing a scalable, empirically evaluable paradigm for intelligent, standards-compliant medical content generation.

3 citationsRead paper

Evaluating Generalization Capabilities of LLM-Based Agents in Mixed-Motive Scenarios Using Concordia

Dec 03, 2025

Existing evaluation methods inadequately measure large language model (LLM) agents’ cooperative generalization capability in novel, mixed-motive social scenarios. Method: We conduct a systematic zero-shot evaluation on the Concordia multi-agent simulation platform, assessing LLM agents’ ability to recognize and realize mutual benefit across diverse social interaction tasks—including negotiation and collective action—using a novel quantitative framework for general cooperative intelligence. This framework emphasizes high-generalization dimensions such as persuasion and norm enforcement. Contribution/Results: Empirical analysis of NeurIPS 2024 Concordia Competition data reveals substantial limitations in current LLM agents’ cross-context cooperative generalization, particularly in dynamic coordination and implicit norm modeling. Our work establishes a new paradigm for benchmarking and diagnosing cooperative intelligence, advancing both methodological rigor and diagnostic precision in multi-agent cooperation research.

2 citationsRead paper
Recent publications

Latest Papers

Offline Reinforcement Learning for Hemodynamic Management of Sepsis in the ICU: a MIMIC-IV Study with Dual Off-Policy Evaluation

Aug 17, 2026

This study addresses the challenges of decision-making uncertainty and unreliable offline evaluation in sepsis ICU treatment. Leveraging the MIMIC-IV dataset, we propose a transparent validation framework integrating doubly robust off-policy estimation, reliability diagnostics, and physician concordance analysis. By combining discrete Markov Decision Processes with Fitted Q-Evaluation to optimize hemodynamic management strategies, this approach effectively mitigates evaluation fragility. Experimental results demonstrate that the learned policy yields significantly higher expected returns than clinician behavior while maintaining low behavioral deviation, thereby balancing improved efficacy with clinical plausibility. These findings provide empirical evidence and methodological innovations for developing safe and reliable AI-driven clinical decision support systems.

0 citationsRead paper