arXiv subjects

Salvatore Greco

Publications and source records attributed to Salvatore Greco.

At least 19 recordsLinked to original sources

arXiv2026

Preference-Based Estimation of Achievement Scalarising Function Parameters via Ordinal Regression

Estimating the parameters of an achievement scalarising function is a challenging task in multiobjective optimisation. Achievement scalarising functions are widely used in several contexts, including preference-based optimisation methods and algorithmic procedures for generating or selecting efficient solutions. However, their effectiveness strongly depends on the appropriate specification of their parameters, such as weights and reference points. In this paper, we propose an ordinal-regression-based methodology to estimate the parameters of an achievement scalarising function from preference information provided by the DM. In particular, the proposed approach allows us to infer both the weights associated with the objectives and the reference points characterising the scalarising function. The methodology is illustrated through a didactic example based on a multiobjective knapsack problem, which shows step by step how the preference information is translated into model parameters. The example illustrates that ordinal regression provides a flexible and interpretable framework for calibrating achievement scalarising functions and supporting the elicitation of decision-maker preferences in multiobjective optimisation.

math.OC↗

arXiv2026

Staying with the Uncertainty: Uncertainty-Scaffolding Strategies for Artificial Moral Advisors in LLM-to-LLM Simulated Conversations

LLMs are increasingly deployed as Artificial Moral Advisors (AMA) in a variety of contexts: what kind of conversational patterns should they display? In this paper, we study how AMA can help their interlocutors "stay with the uncertainty". We propose three modes of uncertainty (Perspective-Multiplying, Tension-Preserving, Process-Reflecting) and compare them against three control conditions (Baseline, Persuasive, Sycophantic). A user-agent LLM engages in a dialogue on an ethical dilemma with an AMA following a specific uncertainty strategy, and completes pre- and post-conversation questionnaires. We further examine the effect of two persona prompt formats (Declarative and Narrative). We found that (1) no single model dominates as a simulated user agent, with open models aligning with human ambiguity through between-persona divergence and closed models through within-persona hedging; (2) declarative personas better capture initial stance diversity while narrative personas show more realistic belief revision; (3) all six AMA strategies produce distinguishable conversational patterns; and (4) uncertainty strategies differ not in how much stance revision they produce, but in the quality of engagement they sustain.

cs.CL↗

arXiv2026

PREF-XAI: Preference-Based Personalized Rule Explanations of Black-Box Machine Learning Models

Explainable artificial intelligence (XAI) has predominantly focused on generating model-centric explanations that approximate the behavior of black-box models. However, such explanations often overlook a fundamental aspect of interpretability: different users require different explanations depending on their goals, preferences, and cognitive constraints. Although recent work has explored user-centric and personalized explanations, most existing approaches rely on heuristic adaptations or implicit user modeling, lacking a principled framework for representing and learning individual preferences. In this paper, we consider Preference-Based Explainable Artificial Intelligence (PREF-XAI), a novel perspective that reframes explanation as a preference-driven decision problem. Within PREF-XAI, explanations are not treated as fixed outputs, but as alternatives to be evaluated and selected according to user-specific criteria. In the PREF-XAI perspective, here we propose a methodology that combines rule-based explanations with formal preference learning. User preferences are elicited through a ranking of a small set of candidate explanations and modeled via an additive utility function inferred using robust ordinal regression. Experimental results on real-world datasets show that PREF-XAI can accurately reconstruct user preferences from limited feedback, identify highly relevant explanations, and discover novel explanatory rules not initially considered by the user. Beyond the proposed methodology, this work establishes a connection between XAI and preference learning, opening new directions for interactive and adaptive explanation systems.

cs.LG↗

arXiv2026

KCLarity at SemEval-2026 Task 6: Encoder and Zero-Shot Approaches to Political Evasion Detection

This paper describes the KCLarity team's participation in CLARITY, a shared task at SemEval 2026 on classifying ambiguity and evasion techniques in political discourse. We investigate two modelling formulations: (i) directly predicting the clarity label, and (ii) predicting the evasion label and deriving clarity through the task taxonomy hierarchy. We further explore several auxiliary training variants and evaluate decoder-only models in a zero-shot setting under the evasion-first formulation. Overall, the two formulations yield comparable performance. Among encoder-based models, RoBERTa-large achieves the strongest results on the public test set, while zero-shot GPT-5.2 generalises better on the hidden evaluation set.

cs.CL↗

arXiv2025

An Explainable and Interpretable Composite Indicator Based on Decision Rules

Composite indicators are widely used to score or classify units evaluated on multiple criteria. Their construction typically involves aggregating criteria evaluations, a common practice in Multiple Criteria Decision Aiding (MCDA). Beyond producing a final score or classification, however, ensuring explainability, interpretability, and transparency is crucial. This paper proposes a novel framework for constructing explainable and interpretable composite indicators using if-then decision rules. We explore four scenarios: (i) decision rules explaining classifications derived from the sum of ordinal indicator codes; (ii) interpretation of an opaque numerical composite indicator used to classify units into quantiles; (iii) construction of a composite indicator from decision-maker preference information, given as classifications of reference units; and (iv) explanation of classifications generated by an existing MCDA method. To induce the rules from scored or classified units, we apply the Dominance-based Rough Set Approach. The resulting rules relate class assignments or scores to threshold conditions on indicator values in a clear and intelligible way, clarifying the underlying rationale and supporting the assessment of new units. Our main methodological contribution is the introduction of a decision-rule-based framework for constructing composite indicators. Moreover, the framework extends naturally to continuous composite indicators by treating each distinct score as an ordered class. This is enabled by a new algorithm that efficiently induces all minimal rules in a single run. Although this may yield many rules, explainability is preserved by showing only those satisfied by the unit of interest. Finally, the methodology can handle datasets with missing values, enhancing its practical applicability.

cs.LG↗

arXiv2025

Optimising the attribute order in Fuzzy Rough Rule Induction

Interpretability is the next pivotal frontier in machine learning research. In the pursuit of glass box models - as opposed to black box models, like random forests or neural networks - rule induction algorithms are a logical and promising avenue, as the rules can easily be understood by humans. In our previous work, we introduced FRRI, a novel rule induction algorithm based on fuzzy rough set theory. We demonstrated experimentally that FRRI outperformed other rule induction methods with regards to accuracy and number of rules. FRRI leverages a fuzzy indiscernibility relation to partition the data space into fuzzy granules, which are then combined into a minimal covering set of rules. This indiscernibility relation is constructed by removing attributes from rules in a greedy way. This raises the question: does the order of the attributes matter? In this paper, we show that optimising only the order of attributes using known methods from fuzzy rough set theory and classical machine learning does not improve the performance of FRRI on multiple metrics. However, removing a small number of attributes using fuzzy rough feature selection during this step positively affects balanced accuracy and the average rule length.

cs.AI↗

arXiv2025

A Multi-Objective Portfolio of Portfolios Problem with Qualitative Performance Assessments

We present a multi-objective portfolio decision model that involves selecting both a portfolio of projects and a set of elements to allocate to each project. Our model includes a defined set of objectives to optimize, with projects contributing to these objectives in various ways. The elements included in the portfolios are assessed based on both qualitative and quantitative criteria. Projects can only be selected for the portfolio if they meet specific requirements defined by threshold values on the criteria. The model is adaptable to include temporal considerations and stochastic, making it suitable for a wide range of real-life applications. To manage the decision-making process, we employ an interactive multi-objective method that integrates the selection of both portfolios and elements. After making initial selections, we ask the decision-maker to evaluate the portfolios, from which we derive a series of rules to be incorporated into the multiobjective model until the decision-maker is satisfied. We illustrate the functionality of our model through an illustrative case study.

math.CO↗

arXiv2024

FRRI: a novel algorithm for fuzzy-rough rule induction

Interpretability is the next frontier in machine learning research. In the search for white box models - as opposed to black box models, like random forests or neural networks - rule induction algorithms are a logical and promising option, since the rules can easily be understood by humans. Fuzzy and rough set theory have been successfully applied to this archetype, almost always separately. As both approaches to rule induction involve granular computing based on the concept of equivalence classes, it is natural to combine them. The QuickRules\cite{JensenCornelis2009} algorithm was a first attempt at using fuzzy rough set theory for rule induction. It is based on QuickReduct, a greedy algorithm for building decision reducts. QuickRules already showed an improvement over other rule induction methods. However, to evaluate the full potential of a fuzzy rough rule induction algorithm, one needs to start from the foundations. In this paper, we introduce a novel rule induction algorithm called Fuzzy Rough Rule Induction (FRRI). We provide background and explain the workings of our algorithm. Furthermore, we perform a computational experiment to evaluate the performance of our algorithm and compare it to other state-of-the-art rule induction approaches. We find that our algorithm is more accurate while creating small rulesets consisting of relatively short rules. We end the paper by outlining some directions for future work.

cs.LG↗

arXiv2024

NLPGuard: A Framework for Mitigating the Use of Protected Attributes by NLP Classifiers

AI regulations are expected to prohibit machine learning models from using sensitive attributes during training. However, the latest Natural Language Processing (NLP) classifiers, which rely on deep learning, operate as black-box systems, complicating the detection and remediation of such misuse. Traditional bias mitigation methods in NLP aim for comparable performance across different groups based on attributes like gender or race but fail to address the underlying issue of reliance on protected attributes. To partly fix that, we introduce NLPGuard, a framework for mitigating the reliance on protected attributes in NLP classifiers. NLPGuard takes an unlabeled dataset, an existing NLP classifier, and its training data as input, producing a modified training dataset that significantly reduces dependence on protected attributes without compromising accuracy. NLPGuard is applied to three classification tasks: identifying toxic language, sentiment analysis, and occupation classification. Our evaluation shows that current NLP classifiers heavily depend on protected attributes, with up to $23\%$ of the most predictive words associated with these attributes. However, NLPGuard effectively reduces this reliance by up to $79\%$, while slightly improving accuracy.

cs.CL↗

arXiv2024

Unsupervised Concept Drift Detection from Deep Learning Representations in Real-time

Concept drift is the phenomenon in which the underlying data distributions and statistical properties of a target domain change over time, leading to a degradation in model performance. Consequently, production models require continuous drift detection monitoring. Most drift detection methods to date are supervised, relying on ground-truth labels. However, they are inapplicable in many real-world scenarios, as true labels are often unavailable. Although recent efforts have proposed unsupervised drift detectors, many lack the accuracy required for reliable detection or are too computationally intensive for real-time use in high-dimensional, large-scale production environments. Moreover, they often fail to characterize or explain drift effectively. To address these limitations, we propose \textsc{DriftLens}, an unsupervised framework for real-time concept drift detection and characterization. Designed for deep learning classifiers handling unstructured data, \textsc{DriftLens} leverages distribution distances in deep learning representations to enable efficient and accurate detection. Additionally, it characterizes drift by analyzing and explaining its impact on each label. Our evaluation across classifiers and data-types demonstrates that \textsc{DriftLens} (i) outperforms previous methods in detecting drift in 15/17 use cases; (ii) runs at least 5 times faster; (iii) produces drift curves that align closely with actual drift (correlation $\geq\!0.85$); (iv) effectively identifies representative drift samples as explanations.

cs.LG↗

arXiv2024

Representation of preferences for multiple criteria decision aiding in a new seven-valued logic

The seven-valued logic considered in this paper naturally arises within the rough set framework, allowing to distinguish vagueness due to imprecision from ambiguity due to coarseness. Recently, we discussed its utility for reasoning about data describing multi-attribute classification of objects. We also showed that this logic contains, as a particular case, the celebrated Belnap four-valued logic. Here, we present how the seven-valued logic, as well as the other logics that derive from it, can be used to represent preferences in the domain of Multiple Criteria Decision Aiding (MCDA). In particular, we propose new forms of outranking and value function preference models that aggregate multiple criteria taking into account imperfect preference information. We demonstrate that our approach effectively addresses common challenges in preference modeling for MCDA, such as uncertainty, imprecision, and ill-determination of performances and preferences. To this end, we present a specific procedure to construct a seven-valued preference relation and use it to define recommendations that consider robustness concerns by utilizing multiple outranking or value functions representing the decision maker s preferences. Moreover, we discuss the main properties of the proposed seven-valued preference structure and compare it with current approaches in MCDA, such as ordinal regression, robust ordinal regression, stochastic multiattribute acceptability analysis, stochastic ordinal regression, and so on. We illustrate and discuss the application of our approach using a didactic example. Finally, we propose directions for future research and potential applications of the proposed methodology.

cs.AI↗

arXiv2024

Deck of Cards method for Hierarchical, Robust and Stochastic Ordinal Regression

We consider the recently introduced application of the Deck of Cards Method (DCM) to ordinal regression proposing two extensions related to two main research trends in Multiple Criteria Decision Aiding, namely scaling and ordinal regression generalizations. On the one hand, procedures, different from DCM (e.g. AHP, BWM, MACBETH) to collect and elaborate Decision Maker's (DM's) preference information are considered to define an overall evaluation of reference alternatives. On the other hand, Robust Ordinal Regression and Stochastic Multicriteria Acceptability Analysis are used to offer the DM more detailed and realistic decision-support outcomes. More precisely, we take into account preference imprecision and indetermination through a set of admissible comprehensive evaluations of alternatives provided by the whole set of value functions compatible with DM's preference information rather than the univocal assessment obtained from a single value function. In addition, we also consider alternatives evaluated on a set of criteria hierarchically structured. The methodology we propose allows the DM to provide precise or imprecise information at different levels of the hierarchy of criteria. Like scaling procedures, the compatible value function we consider can be of a different nature, such as weighted sum, linear or general monotone value function, or Choquet integral. Consequently, the approach we propose is versatile and well-equipped to be adapted to DM's characteristics and requirements. The applicability of the proposed methodology is shown by a didactic example based on a large ongoing research project in which Italian regions are evaluated on criteria representing Circular Economy, Innovation-Driven Development and Smart Specialization Strategies.

math.OC↗

arXiv2024

Operational Research: Methods and Applications

Throughout its history, Operational Research has evolved to include a variety of methods, models and algorithms that have been applied to a diverse and wide range of contexts. This encyclopedic article consists of two main sections: methods and applications. The first aims to summarise the up-to-date knowledge and provide an overview of the state-of-the-art methods and key developments in the various subdomains of the field. The second offers a wide-ranging list of areas where Operational Research has been applied. The article is meant to be read in a nonlinear fashion. It should be used as a point of reference or first-port-of-call for a diverse pool of readers: academics, researchers, students, and practitioners. The entries within the methods and applications sections are presented in alphabetical order. The authors dedicate this paper to the 2023 Turkey/Syria earthquake victims. We sincerely hope that advances in OR will play a role towards minimising the pain and suffering caused by this and future catastrophes.

math.OC↗

arXiv2023

Pawlak, Belnap and the magical number seven

We are considering the algebraic structure of the Pawlak-Brouwer-Zadeh lattice to distinguish vagueness due to imprecision from ambiguity due to coarseness. We show that a general class of many-valued logics useful for reasoning about data emerges from this context. All these logics can be obtained from a very general seven-valued logic which, interestingly enough, corresponds to a reasoning system developed by Jaina philosophers four centuries BC. In particular, we show how the celebrated Belnap four-valued logic can be obtained from the very general seven-valued logic based on the Pawlak-Brouwer-Zadeh lattice.

math.LO↗

arXiv2023

Discounting and Impatience

Understanding how people actually trade off time for money is perhaps the major question in the field of time discounting. There is indeed a vast body of work devoted to explore the underlying mechanisms of the individual decision making process in an intertemporal context. This paper presents a family of new discount functions whereof we derive a formal axiomatization. Applying the framework proposed by Bleichrodt, Rohde and Wakker, we further extend their formulation of CADI and CRDI functions, making discounting a function not only of time delay but, simultaneously, also of time distortion. Our main purpose is, in practice, to provide a tractable setting within which individual intertemporal preferences can be outlined. Furthermore, we apply our models to study the relation between individual time preferences and personality traits. For the CADI-CADI, results show that the habit of smoking is heavily related with both impatience and time perception. Within the Big-Five framework, conscientiousness, agreeableness and openness are positively related with patience (low r, initial discount rate).

econ.GN↗

arXiv2023

A new Ordinal Regression procedure for Multiple Criteria Decision Aiding: the case of the space time model for a sustainable Ecovillage

In this paper, we present a methodology based on a multiobjective optimization suggesting which facility to implement, in which location, and at which time. In this context, we define a new elicitation procedure to handle Decision Makers (DMs) preferences with an intrinsic and more general interest that goes beyond the specific decision problem. In particular, the user's preferences are elicited by conjugating the deck of cards method with the ordinal regression approach allowing the DM to provide preference information in terms of ranking and pairwise comparing with regard to the intensity of preference of some solutions of the optimization problem. Then, the score of the reference solutions obtained through the deck of the cards method is used as a basis for an ordinal regression procedure that, to take into account interaction between criteria, represents DM's multicriteria preferences by means of a value function expressed in terms of a Choquet Integral. The obtained value function is then used to define a multiobjective optimization problem. The new feasible solutions obtained by the resolution of the optimization problem are proposed to the DM to verify his appreciation and collect further new preference information to iterate the interaction procedure ending when the DM is satisfied of the proposed solution. We apply our methodology to a real world problem to handle the planning procedure of a sustainable Ecovillage in the province of Turin (Italy). We consider a set of facilities to be distributed in a given space in a proper temporal sequence that we conveniently formulated in terms of the space time model introduced by Barbati et al. (2020). We interact with the President of the cooperative owning the Ecovillage to detail what facilities of the Ecovillage should be selected among the proposed ones, where they should be located, and when they should be planned.

math.OC↗

arXiv2023

Inducing a probability distribution in Stochastic Multicriteria Acceptability Analysis

In multiple criteria decision aiding, very often the alternatives are compared by means of a value function compatible with the preferences expressed by the Decision Maker. The problem is that, in general, there is a plurality of compatible value functions, and providing a final recommendation on the problem at hand considering only one of them could be considered arbitrary to some extent. For such a reason, Stochastic Multicriteria Acceptability Analysis gives information in statistical terms by taking into account a sample of models compatible with the provided preferences. These statistics are given assuming the existence of a probability distribution in the space of value functions being defined a priori. In this paper, we propose some methods aiming to build a probability distribution on the space of value functions considering the preference information given by the Decision Maker. To prove the goodness of our proposal we performed an extensive set of simulations. Moreover, a sensitivity analysis on the variables of our procedure has been done as well.

math.OC↗

arXiv2022

Dominance-based Rough Set Approach, basic ideas and main trends

Dominance-based Rough Approach (DRSA) has been proposed as a machine learning and knowledge discovery methodology to handle Multiple Criteria Decision Aiding (MCDA). Due to its capacity of asking the decision maker (DM) for simple preference information and supplying easily understandable and explainable recommendations, DRSA gained much interest during the years and it is now one of the most appreciated MCDA approaches. In fact, it has been applied also beyond MCDA domain, as a general knowledge discovery and data mining methodology for the analysis of monotonic (and also non-monotonic) data. In this contribution, we recall the basic principles and the main concepts of DRSA, with a general overview of its developments and software. We present also a historical reconstruction of the genesis of the methodology, with a specific focus on the contribution of Roman Słowiński.

cs.AI↗