Angel Cuenca-Ortega
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Angel Cuenca-Ortega has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Santiago Escobar
VRAIN, Universitat Politècnica de València
6 shared publications - María Alpuente
Universitat Politècnica de València
4 shared publications José Meseguer
University of Illinois at Urbana-Champaign
3 shared publications- Julia Sapiña
VRAIN, Universitat Politècnica de València
2 shared publications María Alpuente
Universitat Politècnica de València
2 shared publications- María Fernanda Molina
Universidad Nacional de Tres de Febrero (UNTREF), Argentina
2 shared publications
Latest Publications
- Open Access
- 21 October 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (10) , 903
Analysis of Public Tenders with Rule-Based Systems to Improve Decision-Making
We evaluate supervised classification models on European Union public tenders, pairing random forest and gradient boosting with explainable artificial intelligence (XAI) and a rule-based system. Using TED data processed with Apache Spark, we modeled the main activity of the contracting authority among 21 classes (Experiment A) and whether an SME wins the contract (Experiment B). Random forest outperforms gradient boosting and a decision tree baseline in every fold of both experiments (one-sided Wilcoxon test, p = 0.001). Its F1-macro in Experiment A is 0.3550, far above the majority-class baseline (0.0228), yet only three of the 21 classes exceed F1 = 0.6, and the score falls to 0.2203 on later years and 0.1443 on countries absent from training. Experiment B is easier: F1-macro reaches 0.5863 (AUC = 0.7100) against a baseline of 0.4500 and 0.5515 when the model sees only the features that are known before bidding. That restricted model, however, has no discriminative ability on unseen countries (AUC = 0.51). National markets explain much of this. In the seven largest, F1-macro averages 0.59 when the model is trained within the country and 0.50 when the country is left out. The if–then rules derived from LIME have low fidelity (R2 = 0.43 and 0.16), and, on held-out data, a third of the Experiment B rules are no more precise than an empty rule; TreeSHAP supports the global feature ranking but not these local explanations. Any use for SME decision support would be limited to markets that are represented in training and would first need validation with end users.
- 21 October 2026
- Published by MDPI AG in Algorithms
- Vol. 19 (10) , 903
Open AccessAnalysis of Public Tenders with Rule-Based Systems to Improve Decision-Making
We evaluate supervised classification models on European Union public tenders, pairing random forest and gradient boosting with explainable artificial intelligence (XAI) and a rule-based system. Using TED data processed with Apache Spark, we modeled the main activity of the contracting authority among 21 classes (Experiment A) and whether an SME wins the contract (Experiment B). Random forest outperforms gradient boosting and a decision tree baseline in every fold of both experiments (one-sided Wilcoxon test, p = 0.001). Its F1-macro in Experiment A is 0.3550, far above the majority-class baseline (0.0228), yet only three of the 21 classes exceed F1 = 0.6, and the score falls to 0.2203 on later years and 0.1443 on countries absent from training. Experiment B is easier: F1-macro reaches 0.5863 (AUC = 0.7100) against a baseline of 0.4500 and 0.5515 when the model sees only the features that are known before bidding. That restricted model, however, has no discriminative ability on unseen countries (AUC = 0.51). National markets explain much of this. In the seven largest, F1-macro averages 0.59 when the model is trained within the country and 0.50 when the country is left out. The if–then rules derived from LIME have low fidelity (R2 = 0.43 and 0.16), and, on held-out data, a third of the Experiment B rules are no more precise than an empty rule; TreeSHAP supports the global feature ranking but not these local explanations. Any use for SME decision support would be limited to markets that are represented in training and would first need validation with end users.
- Open Access
- 10 February 2025
- Published by Universidad Tecnologica Israel in REVISTA ODIGOS
- Vol. 6 (1) , 83-100
Análisis de licitaciones públicas en Ecuador: aplicación de técnicas de explicabilidad en modelos de aprendizaje automático
Las licitaciones públicas permiten a las instituciones contratar bienes, obras y servicios esenciales para el crecimiento del país. Este trabajo consistió en analizar los procesos de licitaciones públicas en Ecuador mediante la aplicación de técnicas de aprendizaje automático y explicabilidad de modelos, con el fin de mejorar la toma de decisiones. Se recopiló y procesó un conjunto de datos abiertos extraídos de la base de datos del SERCOP para identificar patrones y variables clave que influyeron en el éxito de las licitaciones. Utilizando modelos de clasificación como Random Forest, AdaBoost y CatBoost, y técnicas de explicabilidad como SHAP y Feature Importance, se desarrollaron modelos predictivos que permiten comprender de manera transparente las decisiones generadas por los algoritmos. Los resultados mostraron que CatBoost fue el modelo con mayor precisión predictiva y Feature Importance resultó ser la técnica más efectiva para explicar las predicciones. Además, se procedió a crear una interfaz web que permitió ingresar los datos de entrada y determinar si es recomendable que una empresa participe en una licitación. La inteligencia artificial explicable no solo mejora la precisión, sino que también proporciona información valiosa para que las empresas optimicen su participación en estos procesos.
- 10 February 2025
- Published by Universidad Tecnologica Israel in REVISTA ODIGOS
- Vol. 6 (1) , 83-100
Open AccessAnálisis de licitaciones públicas en Ecuador: aplicación de técnicas de explicabilidad en modelos de aprendizaje automático
Las licitaciones públicas permiten a las instituciones contratar bienes, obras y servicios esenciales para el crecimiento del país. Este trabajo consistió en analizar los procesos de licitaciones públicas en Ecuador mediante la aplicación de técnicas de aprendizaje automático y explicabilidad de modelos, con el fin de mejorar la toma de decisiones. Se recopiló y procesó un conjunto de datos abiertos extraídos de la base de datos del SERCOP para identificar patrones y variables clave que influyeron en el éxito de las licitaciones. Utilizando modelos de clasificación como Random Forest, AdaBoost y CatBoost, y técnicas de explicabilidad como SHAP y Feature Importance, se desarrollaron modelos predictivos que permiten comprender de manera transparente las decisiones generadas por los algoritmos. Los resultados mostraron que CatBoost fue el modelo con mayor precisión predictiva y Feature Importance resultó ser la técnica más efectiva para explicar las predicciones. Además, se procedió a crear una interfaz web que permitió ingresar los datos de entrada y determinar si es recomendable que una empresa participe en una licitación. La inteligencia artificial explicable no solo mejora la precisión, sino que también proporciona información valiosa para que las empresas optimicen su participación en estos procesos.
- 23 February 2019
- Published by Springer Nature
- p. 38-55
Homeomorphic Embedding Modulo Combinations of Associativity and Commutativity Axioms
- 23 February 2019
- Published by Springer Nature
- p. 38-55
Homeomorphic Embedding Modulo Combinations of Associativity and Commutativity Axioms
- 30 December 2018
- Published by Springer Nature
- p. 353-368
Educational Software Development in Ecuadorian Universities: A Systematic Mapping Study
- 30 December 2018
- Published by Springer Nature
- p. 353-368
Educational Software Development in Ecuadorian Universities: A Systematic Mapping Study
- Open Access
Homeomorphic Embedding modulo Combinations of Associativity and Commutativity Axioms
The Homeomorphic Embedding relation has been amply used for defining termination criteria of symbolic methods for program analysis, transformation, and verification. However, homeomorphic embedding has never been investigated in the context of order-sorted rewrite theories that support symbolic execution methods modulo equational axioms. This paper generalizes the symbolic homeomorphic embedding relation to order-sorted rewrite theories that may contain various combinations of associativity and/or commutativity axioms for different binary operators. We systematically measure the performance of increasingly efficient formulations of the homeomorphic embedding relation modulo associativity and commutativity axioms. From our experimental results, we conclude that our most efficient version indeed pays off in practice.
Open AccessHomeomorphic Embedding modulo Combinations of Associativity and Commutativity Axioms
The Homeomorphic Embedding relation has been amply used for defining termination criteria of symbolic methods for program analysis, transformation, and verification. However, homeomorphic embedding has never been investigated in the context of order-sorted rewrite theories that support symbolic execution methods modulo equational axioms. This paper generalizes the symbolic homeomorphic embedding relation to order-sorted rewrite theories that may contain various combinations of associativity and/or commutativity axioms for different binary operators. We systematically measure the performance of increasingly efficient formulations of the homeomorphic embedding relation modulo associativity and commutativity axioms. From our experimental results, we conclude that our most efficient version indeed pays off in practice.
- 24 August 2017
- Published by Cambridge University Press (CUP) in Theory and Practice of Logic Programming
- Vol. 17 (5-6) , 689-707
Inspecting Maude variants withGLINTS
This paper introducesGLINTS, a graphical tool for exploring variant narrowing computations in Maude. The most recent version of Maude, version 2.7.1, provides quite sophisticated unification features, including order-sorted equational unification for convergent theories modulo axioms such as associativity, commutativity, and identity. This novel equational unification relies on built-in generation of the set ofvariantsof a termt, i.e., the canonical form oftσ for a computed substitution σ. Variant generation relies on a novel narrowing strategy calledfolding variant narrowingthat opens up new applications in formal reasoning, theorem proving, testing, protocol analysis, and model checking, especially when the theory satisfies thefinite variant property, i.e., there is a finite number of most general variants for every term in the theory. However, variant narrowing computations can be extremely involved and are simply presented in text format by Maude, often being too heavy to be debugged or even understood. TheGLINTSsystem provides support for (i) determining whether a given theory satisfies the finite variant property, (ii) thoroughly exploring variant narrowing computations, (iii) automatic checking of nodeembeddingandclosednessmodulo axioms, and (iv) querying and inspecting selected parts of the variant trees.
- 24 August 2017
- Published by Cambridge University Press (CUP) in Theory and Practice of Logic Programming
- Vol. 17 (5-6) , 689-707
Inspecting Maude variants withGLINTS
This paper introducesGLINTS, a graphical tool for exploring variant narrowing computations in Maude. The most recent version of Maude, version 2.7.1, provides quite sophisticated unification features, including order-sorted equational unification for convergent theories modulo axioms such as associativity, commutativity, and identity. This novel equational unification relies on built-in generation of the set ofvariantsof a termt, i.e., the canonical form oftσ for a computed substitution σ. Variant generation relies on a novel narrowing strategy calledfolding variant narrowingthat opens up new applications in formal reasoning, theorem proving, testing, protocol analysis, and model checking, especially when the theory satisfies thefinite variant property, i.e., there is a finite number of most general variants for every term in the theory. However, variant narrowing computations can be extremely involved and are simply presented in text format by Maude, often being too heavy to be debugged or even understood. TheGLINTSsystem provides support for (i) determining whether a given theory satisfies the finite variant property, (ii) thoroughly exploring variant narrowing computations, (iii) automatic checking of nodeembeddingandclosednessmodulo axioms, and (iv) querying and inspecting selected parts of the variant trees.
- 25 July 2017
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 3-20
Partial Evaluation of Order-Sorted Equational Programs Modulo Axioms
- 25 July 2017
- Published by Springer Nature in Lecture Notes in Computer Science
- p. 3-20
Partial Evaluation of Order-Sorted Equational Programs Modulo Axioms
- Open Access
Inspecting Maude Variants with GLINTS
This paper introduces GLINTS, a graphical tool for exploring variant narrowing computations in Maude. The most recent version of Maude, version 2.7.1, provides quite sophisticated unification features, including order-sorted equational unification for convergent theories modulo axioms such as associativity, commutativity, and identity (ACU). This novel equational unification relies on built-in generation of the set of 'variants' of a term $t$, i.e., the canonical form of $t \sigma$ for a computed substitution $\sigma$. Variant generation relies on a novel narrowing strategy called 'folding variant narrowing' that opens up new applications in formal reasoning, theorem proving, testing, protocol analysis, and model checking, especially when the theory satisfies the 'finite variant property', i.e., there is a finite number of most general variants for every term in the theory. However, variant narrowing computations can be extremely involved and are simply presented in text format by Maude, often being too heavy to be debugged or even understood. The GLINTS system provides support for (i) determining whether a given theory satisfies the finite variant property, (ii) thoroughly exploring variant narrowing computations, (iii) automatic checking of node 'embedding' and 'closedness' modulo axioms, and (iv) querying and inspecting selected parts of the variant trees. This paper is under consideration for acceptance in TPLP.Open AccessInspecting Maude Variants with GLINTS
This paper introduces GLINTS, a graphical tool for exploring variant narrowing computations in Maude. The most recent version of Maude, version 2.7.1, provides quite sophisticated unification features, including order-sorted equational unification for convergent theories modulo axioms such as associativity, commutativity, and identity (ACU). This novel equational unification relies on built-in generation of the set of 'variants' of a term $t$, i.e., the canonical form of $t \sigma$ for a computed substitution $\sigma$. Variant generation relies on a novel narrowing strategy called 'folding variant narrowing' that opens up new applications in formal reasoning, theorem proving, testing, protocol analysis, and model checking, especially when the theory satisfies the 'finite variant property', i.e., there is a finite number of most general variants for every term in the theory. However, variant narrowing computations can be extremely involved and are simply presented in text format by Maude, often being too heavy to be debugged or even understood. The GLINTS system provides support for (i) determining whether a given theory satisfies the finite variant property, (ii) thoroughly exploring variant narrowing computations, (iii) automatic checking of node 'embedding' and 'closedness' modulo axioms, and (iv) querying and inspecting selected parts of the variant trees. This paper is under consideration for acceptance in TPLP.