arXiv subjects

Fabio Palomba

Publications and source records attributed to Fabio Palomba.

At least 19 recordsLinked to original sources

arXiv2026

Specification-Driven Governance for Fairness-by-Design in Agentic Software Engineering: How Far Can We Go?

Agentic AI is shifting developers' responsibilities from software implementation to software governance. While this paradigm improves productivity, it also raises challenges in ensuring that autonomous coding agents preserve developers' intentions. Among these, operationalizing fairness-by-design is largely unexplored, as existing approaches provide limited support for representing and propagating fairness requirements throughout agentic development workflows. In this paper, we investigate whether specification-driven development can operationalize fairness-by-design in the agentic development of machine learning systems. We study three coding agents on four machine learning applications, using four prompts that progressively operationalize fairness through nine Fairness Specification Elements, and analyze the traceability of fairness requirements across the generated artifacts, the fairness-accuracy trade-off of the resulting models, and the agents' design decisions and rationales. The results show that when fairness is operationalized, agents capture almost all fairness requirements in the specification (97.22%) and preserve most of them in the code (82.41%). However, the resulting models are not significantly fairer, as agents implement fairness as a post-hoc audit that neither shapes training nor prevents model delivery. Specifications thus make fairness traceable, but not enforceable, calling for fairness specifications that state what agents should do when requirements are not met, and for workflows that make the fairness decisions agents delegate to developers explicit.

cs.SE↗

arXiv2026

Architecting Fairness-Aware LLM Interactions

Large Language Models (LLMs) are increasingly used as decision-support tools but may generate stereotypical, discriminatory, or unfair responses. As LLMs become embedded in decision-critical software systems, fairness can no longer be treated as a downstream evaluation step, but must become an architectural concern of the interaction itself. Yet current fairness approaches operate as standalone evaluation techniques, lacking a reusable mechanism to verify and mitigate violations at runtime. We propose a fairness-assurance architecture grounded in three reusable design patterns, instantiated as a multi-agent framework that evaluates and iteratively refines LLM responses before delivery. A preliminary evaluation across eight configurations involving different LLM families shows that the architecture identifies fairness violations and iteratively produces compliant responses in all configurations, with behavior differing depending on the model-family composition. These results provide initial evidence for the joint instantiation of the proposed patterns and a first step toward architecting non-functional properties into LLM-based systems.

cs.SE↗

arXiv2026

Reliability-Aware Short-Term Glycemic Risk Prediction Using an Anchor-Based Comparative Framework of Sequence, Multimodal, and State-Updating Digital-Twin Models

Anticipating short-term hypoglycemic and hyperglycemic risk is essential for proactive diabetes management. However, existing machine learning approaches often use inconsistent evaluation protocols and emphasize predictive performance over reliability. This study introduces an anchor-based benchmarking framework integrating longitudinal glucose sequences and multimodal patient information. Using the T1DiabetesGranada dataset, 5,503,441 standardized prediction anchors were constructed from 736 individuals with type 1 diabetes to evaluate hypoglycemia and hyperglycemia prediction at 30-, 60-, and 120-minute horizons. Three neural architectures, a Hybrid LSTM-Transformer, a Multimodal Digital Twin, and a State-Updating Digital Twin with gap-aware longitudinal state progression, were compared under identical patient-level validation protocols. Beyond predictive performance, the evaluation examined calibration, classification thresholds, explanation reliability, cross-model disagreement, and robustness to controlled perturbations. No statistically significant predictive differences were observed after Holm correction, although the models exhibited complementary reliability characteristics. The Multimodal Digital Twin achieved the lowest mean Brier score, while the State-Updating Digital Twin demonstrated the lowest mean expected calibration error (ECE-10). The Hybrid LSTM-Transformer showed strong explanation faithfulness and stability under its temporal attribution protocol. These findings highlight the importance of evaluating digital-twin models beyond predictive accuracy and emphasize the need for reliability assessment and external validation before clinical deployment.

cs.LG↗

arXiv2026

Engineering Requirements for Generative AI Agents: The BARD Process Theory

How teams engineer requirements for systems built around Generative AI (GenAI) Agents remains empirically uncharted. Existing Requirements Engineering (RE) research mainly addresses AI and ML-based systems broadly but provides no process account for agentic systems, whose non-deterministic behavior cannot be anticipated at design time. To understand how teams actually proceed, we applied Socio-Technical Grounded Theory across 36 semi-structured interviews with practitioners, interleaving data collection and analysis so that emerging findings drove our theoretical sampling and shaped successive rounds of coding. The result is Behavioral Agent Requirement Discovery (BARD), the process theory of RE for GenAI Agents. BARD holds that requirements are not fixed up front but discovered through a recurring cycle of four activities: teams scope the need, build the agent, observe its behavior, and revise what they ask for. Through this four-phase loop, the requirement evolves from a fixed specification towards behavioral and evaluation criteria. Four tensions drive transitions between phases, and two conditions frame the loop. BARD provides researchers with a framework for iteratively refining RE workflows, while giving practitioners a structured basis to embed into their development processes.

cs.SE↗

arXiv2026

An Empirical Study of Gemini 3 for Detecting Natural Language Test Smells in Manual Test Cases

Manual testing, in which testers follow natural language instructions to validate system behavior, remains essential for uncovering issues that are difficult to capture with automation. However, manual test cases often contain test smells, quality issues such as ambiguity, redundancy, or missing checks that reduce reliability, maintainability, and reproducibility. Existing detection approaches largely depend on manually engineered rules and thus struggle to generalize and scale across heterogeneous test suites. In our previous work, we assessed the feasibility of using Small Language Models (SLMs) for test smell detection by evaluating GEMMA-3-4B, LLAMA-3.2-3B, and PHI-4-14B on test steps from 143 real-world Ubuntu test cases, covering seven smell types. PHI-4-14B achieved the best performance. In this article, we investigate whether a contemporary Large Language Model (GEMINI-3-PRO-PREVIEW) available at the time of the study can identify test smells in natural language manual test cases using a prompt-based, whole-test-case analysis strategy. Unlike approaches that analyze individual test steps in isolation, our approach evaluates complete test cases, enabling the model to consider relationships and dependencies among test steps. We evaluate the approach on 100 Ubuntu test cases covering seven test smell types and compare its performance against previously evaluated SLMs, including GEMMA-3-4B, LLAMA-3.2-3B, and PHI-4-14B. Our results show that GEMINI-3-PRO-PREVIEW outperforms the SLMs, while producing actionable explanations that can help practitioners revise manual test cases for greater clarity and consistency. We also find that test smells are pervasive in practice, with nearly one detected test smell per step on average, highlighting the need for scalable and automated quality support for manual testing artifacts.

cs.SE↗

arXiv2026

Fairness Hazard Analysis for Socio-Technical Processes: A Multiple-Case Study in Bias-sensitive Organisational Settings

Fairness is increasingly recognised as a first-class requirement in socio-technical processes, where interactions among human actors, software systems, and AI technologies may lead to unfair outcomes in decision-making workflows. If left unaddressed, fairness hazards may accumulate and reinforce systemic bias, highlighting the need to engineer fairness proactively. Despite growing interest in fairness-aware systems, systematic methods for identifying fairness hazards in socio-technical processes and deriving requirements-level mitigations remain limited. To support fairness-by-design during requirements engineering (RE), Fairness Hazard Analysis (FHA) is introduced as a methodology for systematically identifying, analysing, and mitigating fairness hazards. FHA is first assessed through a proof-of-concept validation conducted via two focus groups. Then, a qualitative multiple-case study involving two organisations examines its applicability in real-world settings. The proof-of-concept validation highlighted the benefits derived from the structured nature of the method, and suggested the need to include iterative, dialogic reflection with domain experts. In the multiple case-study where FHA was applied, the practitioners involved were positively impressed by the results and confirmed the relevance of the identified fairness hazards (spanning up to 27% of the process elements), as well as the appropriateness of most of the proposed mitigations, while noting that contextual factors might hinder their implementation. The evaluation also highlighted mitigation patterns, such as independent review and collective decision-making, which can be transferred to different organisations. This paper contributes a structured and empirically validated methodology for integrating fairness considerations in RE and preventing systemic bias in socio-technical processes.

cs.SE↗

arXiv2026

The Language of Security: How Prompt Syntax Shapes Secure Code Generation in Open LLMs

Large Language Models (LLMs) are increasingly used for source code generation despite their outputs often exhibiting security vulnerabilities. Prior work shows that prompt engineering can mitigate such risks, yet (1) they focused on high-level prompting strategies, neglecting recent evidence that fine-grained syntactic variations can substantially alter model behavior; and (2) predominantly evaluate proprietary LLMs, limiting the applicability of their findings in industrial settings where self-hosted, open models are preferred for privacy, compliance, and deployment control. In this paper, we study how fine-grained syntactic constituents of prompts influence the security of open LLM-generated code. Using a parser-driven approach, we systematically generate syntactic variants of security-relevant code generation prompts and evaluate their impact on code security across multiple open LLMs and programming languages. Our results show that specific syntactic elements, such as constraints, guards, conditions, and concept bindings, and their position within the prompt consistently affect the likelihood of generating insecure code. These findings identify prompt syntax as a concrete security control surface and provide actionable guidance for reducing vulnerability risk in LLM-assisted development.

cs.CR↗

arXiv2026

TARS: A Theory-of-Mind Agent for Personalized In-IDE Code Comprehension

Code comprehension is one of the most time-consuming tasks in software engineering, yet most LLM-based assistants produce explanations that ignore who is asking and force developers into a disruptive copy-paste workflow. We present TARS, an LLM-powered agent integrated into Visual Studio Code that supports program comprehension through autonomous explanations anchored directly to the code under analysis. Built around a lightweight Theory of Mind paradigm, TARS profiles a developer's expertise, role, and stylistic preferences, then adapts the depth and tone of its explanations accordingly, grounding them in project documentation via Retrieval-Augmented Generation. To evaluate TARS, we conducted a controlled experiment with 18 participants on non-trivial Java snippets. Participants using TARS completed tasks 26\% faster, reported lower cognitive load, and found the explanations meaningfully adapted to their profiles.

cs.SE↗

arXiv2026

How Do Developers Maintain and Evolve Their Agents' Instructions? An Empirical Study

Context. Autonomous coding agents are increasingly used in software development, shifting parts of the engineering process to AI assistance. While this automation brings clear benefits, it introduces challenges in governance, traceability, and control over agent behavior. Agent Context Files (ACFs) have emerged as a practical mechanism to guide agents through structured instructions, yet little is known about how these artifacts are maintained and how their evolution relates to code development. Objective. This paper plans to investigate the evolution of ACFs and their role in agent-driven development. Specifically, we (1) classify ACF changes through a taxonomy grounded in software maintenance theory, (2) analyze how different types of changes are associated with code quality outcomes, and (3) examine their temporal patterns across the development lifecycle. Method. We conduct a large-scale mining study combining repositories with ACFs and agent-generated commits. We reconstruct ACF evolution at the commit level, classify changes using a qualitative approach, and analyze their association with code quality metrics. Statistical analyses and hypotheses are used to evaluate differences across maintenance categories, to inform future design of ACFs for governing autonomous coding agents.

cs.SE↗

arXiv2026

LLM-Assisted Empirical Software Engineering: Systematic Literature Review and Research Agenda

Context: Empirical Software Engineering (ESE) faces increasing challenges due to data scale, methodological complexity, and reproducibility concerns. Large Language Models (LLMs) have emerged as promising tools to support empirical workflows, yet their use remains fragmented, with no comprehensive synthesis to guide responsible adoption. Aims: This study analyzes how LLMs are used in ESE, examining supported tasks, phases of the empirical lifecycle, integration into workflows, reported benefits and limitations, and the extent of reproducibility-related reporting. It also identifies gaps and future research directions. Method: We conducted a systematic literature review of peer-reviewed papers (2020-2025) across 12 leading software engineering venues, resulting in 50 primary studies analyzed through qualitative and quantitative synthesis. Results: We identified 69 LLM-assisted tasks, mainly in mining software repositories and controlled experiments, focusing on classification, filtering, and evaluation. LLMs are used across multiple phases but are concentrated in data processing and analysis. Their integration is largely automation-oriented, with limited decision-support use. Benefits emphasize efficiency and scalability, while limitations include hallucinations, inconsistency, prompt sensitivity, and reproducibility issues. Reporting practices are often incomplete. Conclusion: LLM use in ESE is growing but remains automation-driven, with gaps in human-centered integration and transparency. We outline implications and research agenda for responsible use.

cs.SE↗

arXiv2026

From Pixels to Explanations: Interpretable Diabetic Retinopathy Grading with CNN-Transformer Ensembles, Visual Explainability and Vision-Language Models

The quality of diabetic retinopathy (DR) screening relies on the ability to correctly grade severity; however, many deep-learning (DL) classifiers cannot be easily interpreted in the clinical context. This study presents a methodology that combines strong discriminative models with multimodal explanations, converting retinal pixels into clinically interpretable outputs. Using the APTOS 2019 benchmark, we evaluated six representative CNN- and transformer-based backbones under a controlled protocol with stratified five-fold cross-validation. We then compared ensembling strategies (hard voting, weighted soft voting, stacking) and investigated a hybrid class-level fusion variant to exploit grade-specific advantages. For interpretability, we produced Grad-CAM++ visual attribution maps and short textual rationales using vision-language models (VLMs) conditioned on the fundus image and classifier outputs under conservative prompting constraints. Modern CNN backbones (ResNet-50 and ConvNeXt-Tiny) provided the strongest single-model baselines, with cross-validated QWK up to 0.919 and 0.914, respectively. Ensembling improved ordinal agreement, and weighted soft voting was the most consistent across folds (QWK 0.934 +/- 0.017). Hybrid class-level fusion was competitive but did not yield a statistically reliable improvement over standard fusion in paired fold comparisons (Holm-adjusted p >= 1.000). For explanation quality, Grad-CAM++ offered plausible but coarse localization, and VLM rationales were generally grade-consistent. Quantitatively, VLM variants showed a trade-off between clinical completeness and template-level semantic similarity (coverage 0.700 vs. BERTScore 0.072), while image-text alignment was comparable (CLIPScore approximately 0.34).

cs.CV↗

arXiv2026

Exploring Individual Factors in the Adoption of LLMs for Specific Software Engineering Purposes

Context: The advent of Large Language Models (LLMs) is transforming software development, significantly enhancing software engineering (SE) processes. Research has explored their role within development teams, focusing on the specific purposes for which LLMs are used within SE tasks, such as artifact generation, decision-making support, and information retrieval. Despite the growing body of work on LLMs in SE, most studies have centered on broad adoption trends, neglecting the nuanced relationship between individual cognitive and behavioral factors and their impact on purpose-specific adoption. While factors such as perceived effort and performance expectancy have been explored at a general level, their influence on distinct SE purposes remains underexamined. This gap hinders the development of tailored LLM-based systems (e.g., Generative AI Agents) that align with engineers' specific needs and limits the ability of team leaders to devise effective strategies for fostering LLM adoption in targeted workflows. Objectives: For the reasons mentioned above, this study aims to study the individual factors that drive the choice to use LLMs for distinct SE purposes. Methods: To achieve the above-mentioned objective, we surveyed 188 software engineers to test the relationship between individual attributes related to technology adoption and LLM adoption across five key purposes, using structural equation modeling (SEM). The Unified Theory of Acceptance and Use of Technology (UTAUT2) was applied to characterize individual adoption behaviors. Results: The findings reveal that purpose-specific adoption is influenced by distinct factors, some of which negatively impact adoption when considered in isolation, underscoring the complexity of LLM integration in SE.

cs.SE↗

arXiv2026

SCOPE: A Dataset of Stereotyped Prompts for Counterfactual Fairness Assessment of LLMs

Large Language Models (LLMs) now serve as the foundation for a wide range of applications, from conversational assistants to decision support tools, making the issue of fairness in their results increasingly important. Previous studies have shown that LLM outputs can shift when prompts reference different demographic groups, even when intent and semantic content remain constant. However, existing resources for probing such disparities rely primarily on small, template-based counterfactual examples or fixed sentence pairs. These benchmarks offer limited linguistic diversity, narrow topical coverage, and little support for analyzing how communicative intent affects model behavior. To address these limitations, we introduce SCOPE (Stereotype-COnditioned Prompts for Evaluation), a large-scale dataset of counterfactual prompt pairs designed to enable systematic investigation of group-sensitive behavior in LLMs. SCOPE contains 241,280 prompts organized into 120,640 counterfactual pairs, each grounded in one of 1,438 topics and spanning nine bias dimensions and 1,536 demographic groups. All prompts are generated under four distinct communicative intents: Question, Recommendation, Direction, and Clarification, ensuring broad coverage of common interaction styles. This resource provides a controlled, semantically aligned, and intent-aware basis for evaluating fairness, robustness, and counterfactual consistency.

cs.SE↗

arXiv2026

Bias Ahead: Sensitive Prompts as Early Warnings for Fairness in Large Language Models

Large Language Models (LLMs) are being increasingly integrated into software systems, offering powerful capabilities but also raising concerns about fairness. Existing fairness benchmarks, however, focus on stereotype-specific associations, which limit their ability to anticipate risks in diverse, real-world contexts. In this paper, we propose sensitive prompts as a new abstraction for fairness evaluation: inputs that are not inherently biased but are more likely to elicit biased or inadequate responses due to the sensitivity of their content. We construct and release SensY, a dataset of 12,801 prompts, categorized as sensitive and non-sensitive, spanning seven thematic domains, combining synthetic generation and real user inputs. Using this dataset, we query three open-source LLMs and manually analyze 4,500 responses to evaluate their adequacy in answering sensitive prompts. Results show that while models often provide factually correct answers, they frequently fail to acknowledge the ethical, relational, or contextual implications of sensitive inputs. In addition, we develop an automated classifier for predicting prompt sensitivity, achieving robust performance on sensitive prompts. Our findings demonstrate that prompt sensitivity can serve as an effective early-warning mechanism for fairness risks in LLMs. This perspective shifts fairness assessment from reactive mitigation toward preventive design, enabling developers to anticipate and manage bias before deployment.

cs.SE↗

arXiv2026

Investigating Technical Debt Types, Issues, and Solutions in Serverless Computing

Serverless computing is a cloud execution model where developers run code, and the server management is handled by the cloud provider. Serverless computing is increasingly gaining popularity as more systems adopt it to enhance scalability and reduce operational costs. While it has numerous benefits, it also embodies unique challenges inherent to serverless computing. One such challenge is Technical Debt (TD), which is exacerbated by the complexities of the serverless paradigm. While prior work has investigated the activities and bad practices that lead to TD in serverless computing, there remains a gap in understanding how TD manifests, the challenges it poses, and the solutions proposed to address TD issues in serverless systems. This study aims to investigate TD in the serverless context using Stack Overflow (SO) as a knowledge base. We collected 78,867 serverless questions on SO and labeled them as TD or non-TD using deep learning. Moreover, we conducted an in-depth analysis to identify types of TD in serverless settings, associated issues, and proposed solutions. We found that 37% of the serverless questions on SO are TD-related. We also identified six serverless-specific issues. Our research highlights the need for tools that can effectively detect TD in serverless applications.

cs.SE↗

arXiv2026

Green Architectural Tactics in ML-enabled Systems: An LLM-based Repository Mining Study

Context: The increasing adoption of machine learning (ML) and artificial intelligence (AI) technologies raises growing concerns about their environmental sustainability. Developing and deploying ML-enabled systems is computationally intensive, particularly during training and inference. Green AI has emerged to address these issues by promoting efficiency without sacrificing accuracy. While prior research has proposed catalogs of sustainable practices (i.e., green tactics), there remains limited understanding of their adoption in practice and whether additional, undocumented tactics exist. Objective: This study aims to investigate the extent to which existing sustainable practices are implemented in real-world ML-enabled systems and to identify previously undocumented practices that support environmental sustainability. Method: We conduct a mining software repository study on 205 open-source ML projects on GitHub. To support our analysis, we design a novel mechanism based on large language models (LLMs) capable of identifying both known and new sustainable practices from code repositories. Results: Our findings confirm that green tactics reported in the literature are used in practice, although adoption rates vary. Furthermore, our LLM-based approach reveals nine previously undocumented sustainable practices. Each tactic is supported with code examples to aid adoption and integration. Conclusions: We finally provide insights for practitioners seeking to reduce the environmental impact of ML-enabled systems and offer a foundation for future research in automating the detection and adoption of sustainable practices.

cs.SE↗

arXiv2026

Socio-Technical Well-Being of Quantum Software Communities: An Overview on Community Smells

Quantum computing has gained significant attention due to its potential to solve computational problems beyond the capabilities of classical computers. With major corporations and academic institutions investing in quantum hardware and software, there has been a rise in the development of quantum-enabled systems, particularly within open-source communities. However, despite the promising nature of quantum technologies, these communities face critical socio-technical challenges, including the emergence of socio-technical anti-patterns known as community smells. These anti-patterns, prevalent in open-source environments, have the potential to negatively impact both product quality and community health by introducing technical debt and amplifying architectural and code smells. Despite the importance of these socio-technical factors, there remains a scarcity of research investigating their influence within quantum open-source communities. This work aims to address this gap by providing a first step in analyzing the socio-technical well-being of quantum communities through a cross-sectional study. By understanding the socio-technical dynamics at play, it is expected that foundational knowledge can be established to mitigate the risks associated with community smells and ensure the long-term sustainability of open-source quantum initiatives.

cs.SE↗

arXiv2026

Tracing Stereotypes in Pre-trained Transformers: From Biased Neurons to Fairer Models

The advent of transformer-based language models has reshaped how AI systems process and generate text. In software engineering (SE), these models now support diverse activities, accelerating automation and decision-making. Yet, evidence shows that these models can reproduce or amplify social biases, raising fairness concerns. Recent work on neuron editing has shown that internal activations in pre-trained transformers can be traced and modified to alter model behavior. Building on the concept of knowledge neurons, neurons that encode factual information, we hypothesize the existence of biased neurons that capture stereotypical associations within pre-trained transformers. To test this hypothesis, we build a dataset of biased relations, i.e., triplets encoding stereotypes across nine bias types, and adapt neuron attribution strategies to trace and suppress biased neurons in BERT models. We then assess the impact of suppression on SE tasks. Our findings show that biased knowledge is localized within small neuron subsets, and suppressing them substantially reduces bias with minimal performance loss. This demonstrates that bias in transformers can be traced and mitigated at the neuron level, offering an interpretable approach to fairness in SE.

cs.SE↗