<?xml version="1.0" encoding="utf-8"?>
    <rss version="2.0">
      <channel xmlns:content="http://purl.org/rss/1.0/modules/content/">
        <title>Frontiers in Computer Science | Software section | New and Recent Articles</title>
        <link>https://www.frontiersin.org/journals/computer-science/sections/software</link>
        <description>RSS Feed for Software section in the Frontiers in Computer Science journal | New and Recent Articles</description>
        <language>en-us</language>
        <generator>Frontiers Feed Generator,version:1</generator>
        <pubDate>2026-11-05T16:29:02.535+00:00</pubDate>
        <ttl>60</ttl>
        <item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1915137</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1915137</link>
        <title><![CDATA[Structurally conditioned counterfactual generation and validation for question answering from scientific text]]></title>
        <pubdate>2026-10-21T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ivan P. Malashin</author><author>Dmitry A. Martysyuk</author><author>Vadim S. Tynchenko</author><author>Andrei P. Gantimurov</author><author>Vladimir A. Nelyub</author><author>Aleksei S. Borodulin</author>
        <description><![CDATA[IntroductionA large language model (LLM)-based pipeline is presented for structured processing of scientific abstracts and for generating paired factual and counterfactual question-answer (QA) instances under document-level intervention constraints.MethodsThe implemented system extracts entities, events, and relations; induces a lightweight structural causal model (SCM) over scientific variables such as method, data, intervention, outcome, and metric; generates five factual QA pairs per abstract; rewrites them under one selected intervention; and filters inconsistent counterfactuals. The source export contains 100 Scopus abstracts, of which 67 documents with complete pipeline artifacts form the validation cohort. Generation uses gpt-4.1, while automated validation uses gpt-5-mini under a 15-criterion checklist; three human annotators independently score the same 67 documents.ResultsHuman agreement is moderate, with a mean pairwise Cohen's kappa of approximately 0.54. Automated validation is high for original QA and relation-level correctness (0.97 and 0.89) but very low for document-level global consistency (0.07), revealing a substantial gap between local correctness and cross-stage coherence. Under the score-based intervention analysis, 25.4% of documents exhibit a validation decrease above the predefined threshold.DiscussionWe refer to this quantity as the intervention degradation rate rather than as bias propagation, because the interventions concern scientific-content variables rather than protected attributes. The study evaluates generation quality, intervention sensitivity, and cross-stage consistency; it does not train a foundation model or claim training-time bias mitigation.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1972880</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1972880</link>
        <title><![CDATA[An integrated AI-based software architecture for load forecasting, controlled anomaly detection, and predictive energy management]]></title>
        <pubdate>2026-10-15T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Karlygash Alibekkyzy</author><author>Gulnur Massabayeva</author><author>Michał Majka</author><author>Madina Bazarova</author><author>Daniyar Kaliyev</author><author>Talshyn Keribayeva</author><author>Saule Belginova</author><author>Alina Bugubayeva</author>
        <description><![CDATA[The increasing deployment of smart meters and Internet of Things technologies has created a demand for intelligent computational systems capable of jointly forecasting electricity consumption, detecting abnormal behavior, and optimizing load management. This study proposes an integrated artificial intelligence-based framework for automated energy metering and consumption management in smart-grid environments. The framework combines a Long Short-Term Memory network for short-term load forecasting, an autoencoder-based model for anomaly detection, and Model Predictive Control for demand-side optimization. The system architecture includes data acquisition, communication, preprocessing, predictive analytics, anomaly monitoring, and decision-support layers. The primary empirical evaluation used 4,380 measured hourly observations (approximately 6 months) from a single administrative building. A separately flagged synthetic temporal extension was retained only as an auxiliary dataset and was excluded from all primary performance metrics. Across five independent model initializations, the final LSTM achieved a MAPE of 3.423 ± 0.555%, an RMSE of 0.336 ± 0.048 kW, and an R2 of 0.973 ± 0.007 on the 633-sequence independent measured test subset. The synthetic extension was maintained as a separately identified auxiliary component and was excluded from all primary training, validation, and independent test metrics. Using a validation-selected reconstruction-error threshold, the autoencoder detected 12 of 30 controlled anomalies in the independent test subset, yielding 100.00% precision, 40.00% recall, and an F1-score of 57.14%. Under the same controlled protocol, One-Class SVM achieved higher recall (53.33%) and F1-score (69.57%), indicating that the autoencoder should be interpreted as an architectural component rather than as the best-performing anomaly detector for the present dataset. A historical proof-of-concept MPC simulation, retained for architectural illustration, reported a 15% reduction in simulated peak demand and a 12% reduction in estimated electricity cost; these values are not interpreted as measured outcomes from the administrative-building experiment. Controlled anomaly injection was used to establish reproducible validation and test labels for the anomaly-detection component. The results demonstrate that the joint integration of forecasting, anomaly detection, and predictive control provides a computationally coherent framework for supporting energy-management efficiency, operational reliability, and data-driven decision-making in intelligent energy systems.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1946723</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1946723</link>
        <title><![CDATA[Constructing t-way test suites: a systematic survey and qualitative synthesis of the One-Parameter-at-a-Time approach]]></title>
        <pubdate>2026-10-15T00:00:00Z</pubdate>
        <category>Review</category>
        <author>Muhammad Sabo Yahaya</author><author>Ahmad Sobri Hashim</author><author>Hussaini Mamman</author><author>Shamsu Abdullahi</author><author>Umama Umer</author><author>Dahiru Adamu Aliyu</author><author>Idrissa Mohammed Godowoli</author>
        <description><![CDATA[Parameter interactions are a major source of software failures, yet the exponential growth of possible configurations makes exhaustive testing impractical. Combinatorial Interaction Testing (CIT) is a highly effective sampling technique for identifying software defects caused by parameter interactions. Over the past two decades, the One-Parameter-at-a-Time (OPAT) constructions paradigm has emerged as a cornerstone of CIT test suite generation. Although numerous OPAT strategies have been proposed to optimize test suite size and execution speed, the literature remains fragmented, and direct side-by-side quantitative benchmarking has been impeded by a lack of open-source tooling and standardized platforms. To address this, this study presents a systematic survey and qualitative synthesis of 40 primary OPAT studies published between 2000 and 2026, retrieved from four major digital libraries. Rather than attempting a direct quantitative benchmark comparison of execution times, which is shown to be mathematically invalid due to highly heterogeneous self-reported hardware environments and the closed-source nature of most academic tools, we establish a unified taxonomy that classifies OPAT research into five major categories: deterministic refinements, scaling and distribution strategies, advanced feature-enhanced strategies, metaheuristic and hybrid strategies, and specialized OPAT adaptations. Our analysis synthesizes qualitative trends, structural characteristics, parameter-order growth mechanisms, and feature-integration gaps across the literature. We conclude with a critical assessment of the tooling and reproducibility challenges in the field, outlining vital future pathways, including diversified metaheuristic and hyper-heuristic techniques, deterministic parallel algorithms, and AI-driven adaptive test generation, to transition OPAT research into industrial-grade, open-source software engineering pipelines.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1960635</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1960635</link>
        <title><![CDATA[TraceRank: configurable review support for requirements traceability recovery]]></title>
        <pubdate>2026-10-14T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Hamad Alsawalqah</author><author>Ahmad Abadleh</author><author>Yazn Alshamaileh</author><author>Marwan Al-Tawil</author><author>Ibrahim Aljarah</author><author>Mohammed Saad Alaydaa</author>
        <description><![CDATA[IntroductionRequirements traceability recovery is commonly evaluated as a single ranking task, although practical review settings differ in the evidence available before review, the inspection budget, and the possibility of analyst feedback. We present TraceRank, a configurable review-support framework that makes recovery assumptions explicit and enables only the recovery modes that are valid for the available evidence and review process.MethodsTraceRank separates cold-start recovery, partial-trace recovery, selective validation, fixed-budget review, and interactive feedback. It organizes retrieval, semantic-structural, dependency, hypergraph, validated-link, and feedback evidence within a common recovery pipeline. We evaluate TraceRank on five traceability datasets: eTour, iTrust, SMOS, eANCI, and LibEST.ResultsIn the cold-start setting, the best retained configuration recovers 404 of 501 true links at Top-30 (LinkRecall@30 = 80.6%) and completes 33 of 60 held-out requirements. Under the partial-trace assumption, the Validated Trace Responsibility Graph (VTRG) recovers 465 links (92.8%), completes 43 requirements, and reduces severe misses to 12. Across the same five requirement-level split seeds used for the split-robustness analysis, VTRG achieves mean LinkRecall@30 of 91.3% (SD 4.0%) and mean CompleteRate@30 of 76.5% (SD 8.1%). Archived fixed-budget and simulated-feedback outputs are retained only as exploratory review-process sensitivity evidence because the originating row-level priority queue and sequential update history are not preserved in the clean archive; they are therefore not treated as confirmatory evidence in the primary conclusions.DiscussionThe results support a configuration-aware interpretation of trace recovery: cold-start, partial-trace, fixed-budget, and feedback-based recovery are different operating settings and should not be interpreted as one universal static ranking problem.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1919950</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1919950</link>
        <title><![CDATA[From model trust to software topology: a Perspective on structurally governed agentic AI systems]]></title>
        <pubdate>2026-10-08T00:00:00Z</pubdate>
        <category>Perspective</category>
        <author>Michai Morin</author><author>Fahim Sufi</author>
        <description><![CDATA[Large language models are increasingly embedded in agentic software systems that retrieve information, plan subtasks, call tools, modify code, execute tests, and adapt through memory or feedback. This transition changes the safety problem: when a model acts through software tools, failure can alter external state. This Perspective argues that agentic AI safety should move beyond exclusive reliance on model trust, alignment, or prompt-level defenses and should also be treated as a software-architecture problem. We propose software topology as a design lens for structurally governed autonomy and develop a five-plane reference topology comprising intent, orchestration, execution, oversight, and adaptation. Its core invariant is that components that decide should not directly act, and components that act should not act without supervision. The topology is realized in Polos, an open reference specification that encodes role contracts, message schemas, declared flow graphs, bounded loops, governed self-improvement, and validator-checked invariants. We argue that this approach complements model alignment and empirical red teaming by providing an inspectable substrate for evaluating, constraining, and governing tool-using AI agents.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1911927</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1911927</link>
        <title><![CDATA[Reusing obsolete windows 10 PCs for on-premises large language model inference]]></title>
        <pubdate>2026-10-08T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ian Curington</author><author>Kevin Lano</author>
        <description><![CDATA[The Windows 10 end-of-life provides a material incentive to repurpose legacy PCs as on-premises large language model (LLM) inference servers. This study assesses the technical feasibility, economic viability and the environmental and social issues of converting an enterprise workstation with an NVIDIA Graphics Processing Unit (GPU) into a dedicated AI node. Using Proxmox virtualisation, Ollama, and Open WebUI to benchmark several open-source LLMs, this study shows that a repurposed workstation, fitted with a single consumer-grade NVIDIA GPU, can deliver stable inference throughput of 29–65 tokens/s for bounded, conversational coding and document-assistant workflows, while preserving full data sovereignty and eliminating recurring subscription fees. Performance analysis reveals that resource contention and GPU memory allocation are the primary determinants of throughput. Correctness, assessed against three established public benchmarks with gold answers, does not track the same ranking as throughput; combining both axes identifies gemma4:e4b as the strongest model evaluated on this hardware for both speed and correctness, while the slowest model tested was among the most accurate. A cost-benefit model shows local operation is substantially cheaper than mid-range and premium cloud alternatives, and only modestly costlier than the cheapest cloud tier on a per-prompt basis—a gap this study argues is outweighed by the digital-sovereignty benefits of local control, particularly where compliance-driven data-handling constraints apply. The study summarizes key background motivations for on-premises AI—privacy, security, regulatory risk—and estimates that this repurposed workstation's usage-phase CO2e footprint is lower than each of the cloud alternatives compared, in addition to avoiding the embodied manufacturing emissions of new hardware. Repurposing legacy workstations with readily available open source tools offers a sustainable, cost-effective pathway for organizations to deploy practical LLM inference capabilities locally, providing a strategic alternative to cloud services for entities that prioritize privacy, regulatory compliance or long-term cost containment.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1928487</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1928487</link>
        <title><![CDATA[A same-condition decision-risk audit for CPU-only machine-learning efficiency measurement]]></title>
        <pubdate>2026-10-06T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Khudran M. Alzhrani</author>
        <description><![CDATA[CPU-only machine-learning efficiency is commonly reported as wall-clock time, throughput, estimated GFLOP/s, or effective bandwidth. This paper introduces a same-condition decision-risk audit that evaluates whether such reported differences are supported by repeated-measurement evidence. The audit constructs two estimates from repeated measurements taken under the same documented condition and quantifies how often their relative difference exceeds a practical relative-difference threshold. We apply the audit to eight CPU-only workload scenarios—dense numerical computation, memory movement, framework-level inference, and single-step training—executed across two local Windows environments and one managed Google Colab CPU allocation. Despite consistent functional-output summaries, timing and derived-metric dispersion depend strongly on workload role, execution context, and phase policy. Under a controlled measurement condition with a fixed phase policy, cross-block same-condition comparisons exceeded a 10% relative-difference threshold in 77.6%–80.6% of cases when reporting a single timing observation, and in 67.1%–76.1% when using the mean of 10 observations, across five representative single-thread scenarios. Approximate sample-size estimates indicate that 81–110 timing observations per group would be needed to support a 10% target relative difference in wall-clock time under the observed variability. The audit demonstrates that CPU-only efficiency differences can only be interpreted meaningfully when the measurement context, phase policy, block structure, same-condition A/A threshold-exceedance, and estimated observation counts are explicitly reported.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1925314</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1925314</link>
        <title><![CDATA[Experimental evaluation of compression algorithms for memory-constrained embedded systems]]></title>
        <pubdate>2026-09-23T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Tomáš Bača</author><author>Lukáš Formanek</author><author>Peter Ševčík</author><author>Ondrej Karpiš</author>
        <description><![CDATA[IntroductionEmbedded systems are constrained by limited flash memory, RAM (random access memory), and processing power, making efficient local data storage and transmission challenging. This study evaluates the suitability of several lossless compression algorithms for resource-constrained embedded platforms and assesses how data representation influences compression performance.MethodsLZW, Heatshrink, LZ4, FastLZ, and Miniz were evaluated directly on an STM32U575ZI microcontroller (Arm Cortex-M33, 160 MHz). A 104,067-byte sensor dataset comprising repetitive time-series measurements of soil and ambient temperature, humidity, battery voltage, timestamps, and identifiers was used as the test workload. The dataset was stored in three formats: CSV (comma-separated values) (104,067 bytes), binary (33,576 bytes), and bit-packed binary (26,581 bytes). Compression ratio, execution time, and total application RAM consumption, including input and output buffers, were measured.ResultsMiniz achieved the highest compression efficiency, reducing the 33,576-byte binary dataset to 11,352 bytes (33.81% of the original size). The fastest compression was obtained with LZ4 configured with acceleration = 16, which compressed the bit-packed representation in 9.437 ms and produced a 25,692-byte output. Among configurations selected for minimum compressed output, FastLZ achieved the shortest execution time at 31.653 ms. When total RAM consumption was evaluated consistently, including all required buffers, LZ4 operating on the bit-packed representation required the least memory among the minimum-output configurations, consuming 58,496 bytes.DiscussionThe results demonstrate significant trade-offs between compression ratio, execution speed, and memory requirements in embedded environments. Data representation substantially influences compression effectiveness, with binary and bit-packed formats generally improving efficiency over CSV. The findings provide practical guidance for selecting compression algorithms in resource-constrained embedded systems, where minimizing storage, execution time, or memory usage may be prioritized depending on application requirements.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1937190</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1937190</link>
        <title><![CDATA[Replications, Revisions, and Reanalyses: Managing Empirical Evidence in Software Engineering]]></title>
        <pubdate>2026-09-22T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Julian Frattini</author><author>Jannik Fischbach</author><author>Davide Fucci</author><author>Michael Unterkalmsteiner</author><author>Daniel Mendez</author><author>Robert Feldt</author><author>Richard Torkar</author>
        <description><![CDATA[BackgroundOne aspired outcome of empirical research on quantitative data is a variance theory, i.e., a quantification of the effect of an independent on a dependent variables. In software engineering (SE) research, variance theories quantify—among others—the impact of tools, techniques, and other treatments on software development outcomes like productivity, cost-efficiency, and defect-detection rates. The validity of variance theories stems from the synthesis of multiple pieces of evidence, which increases its validity beyond the findings of a single study.GapHowever, research synthesis in SE is rare and—if done—mostly limited to purely narrative syntheses. At best, researchers perform meta-analyses to synthesize variance theories from several quantitative results. But even meta-analyses only produce reliable results when synthesizing exact replications yet fail to generalize from variations.GoalWe aim to extend the frontier of research synthesis beyond the state-of-the-art to systematically manage empirical evidence and its evolution.MethodWe apply method engineering to construct a framework for research synthesis from proven, individual method fragments. The framework allows researchers to put new evidence in a clear relation to an existing body of evidence and systematically expand knowledge about a studied phenomenon. We demonstrate the application of this framework to two fields of research by explicitly modeling the relationship between existing pieces of evidence.ResultThe framework puts three types of evolution of evidence into relation: (1) replications investigate the same hypothesis in a new context to improve external validity, (2) revisions challenge an existing hypothesis to improve internal validity, and (3) reanalyses replace analysis methods to improve conclusion validity. Through a systematic evolution of evidence and clear assessment criteria for each dimension of validity, the proposed framework can determine the frontier of a field of research.ConclusionThe framework provides a perspective to systematically evolve empirical evidence in SE, supporting more constructive and productive advances in our field.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1883327</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1883327</link>
        <title><![CDATA[Forecasting sports outcomes through machine learning]]></title>
        <pubdate>2026-09-18T00:00:00Z</pubdate>
        <category>Systematic Review</category>
        <author>Andres Gregori</author><author>Ivan Reyes-Pedroza</author><author>Ari Yair Barrera-Animas</author><author>Julieta Noguez</author><author>David Escobar-Castillejos</author>
        <description><![CDATA[IntroductionMachine learning has increasingly been applied to forecast sports outcomes, assess player performance, estimate injury risk, and model in-game events, yet the reported evidence remains fragmented across sports and validation practices.MethodsTo address this fragmentation, this systematic review followed PRISMA 2020 reporting principles and analyzed 118 English-language journal articles published between 2020 and 2025 in Scopus and Web of Science, coding each study by prediction domain, sport discipline, algorithm family, dataset type, evaluation metric, and validation approach.ResultsMatch outcome prediction and player performance assessment accounted for most studies, whereas injury prediction and in-game event forecasting were less frequent and drew on smaller per-domain samples. Random Forest, Gradient Boosting, XGBoost, Support Vector Machines, Neural Networks, and Logistic Regression were the most common methods: ensemble models predominated in structured tabular datasets, whereas deep learning was concentrated in video, tracking, temporal, and multimodal contexts. Reported performance was difficult to compare directly because studies used heterogeneous metrics, partitions, datasets, and validation schemes.DiscussionThese patterns indicate that data access, task design, and validation quality matter as much as algorithm choice; future work should therefore prioritize open benchmarks, external validation, and causal modeling, while privacy-aware data governance, uncertainty estimation, and interpretability remain underdeveloped priorities for the field.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1813208</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1813208</link>
        <title><![CDATA[A model-driven framework for automated UI prototyping with the two-hemisphere model]]></title>
        <pubdate>2026-09-14T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Kristaps Babris</author><author>Oksana Nikiforova</author><author>Leszek Maciaszek</author>
        <description><![CDATA[IntroductionAutomated generation of executable user interfaces from domain models remains a challenging problem in model-driven software engineering, particularly when both application behavior and domain structure must be preserved in the resulting interface. Existing model-based approaches often remain at the level of abstract interface specifications, whereas AI-assisted generation provides flexible code synthesis but offers limited deterministic traceability to formal domain models. This study proposes a model-driven framework for automatically generating executable web user interface prototypes from the two-hemisphere model, which integrates process and concept representations of a problem domain.MethodsThe proposed transformation pipeline maps processes to application states and navigation structures, concepts and their attributes to data-oriented interface elements, and process semantics to reusable user interface patterns selected through a deterministic scoring mechanism. A proof-of-concept implementation generates React.js/Material-UI prototypes incorporating routing, state management, CRUD structures, and automatically generated demonstration data. The framework was evaluated using an e-commerce case study through three complementary methods: static source-code analysis, ISO/IEC/IEEE 29119-based functional acceptance testing, and a pilot expert heuristic evaluation based on Nielsen's usability heuristics.ResultsThe generated prototype passed all 25 functional acceptance-test scenarios, required no post-generation source-code modification, and was generated in approximately 12 seconds. Static analysis identified no JSLint or ESLint issues and only two minor JSHint warnings. Three independent usability evaluators produced 22 individual observations, which were consolidated into nine unique usability problems: three cosmetic, five minor, one major, and no critical violations. The major issue concerned the absence of a clearly visible order summary in the generated shopping-cart interface.DiscussionThe results provide preliminary evidence that the two-hemisphere model can serve as a viable source representation for generating executable user interface prototypes while preserving traceability between behavioral and structural domain information and the resulting implementation. The heuristic findings also identify concrete opportunities for improving transformation rules and the user interface pattern library. However, broader generalisability, comparative efficiency, accessibility, and end-user usability remain to be established through multi-case and user-centred evaluation.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1886850</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1886850</link>
        <title><![CDATA[Monitoring and comparative analysis of containerized environments]]></title>
        <pubdate>2026-08-31T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Assel Omarbekova</author><author>Lazzat Kussepova</author><author>Altanbek Zulkhazhav</author><author>Mamyr Altaibek</author><author>Gulmira Bekmanova</author><author>Aizhan Nazyrova</author><author>Zhanar Lamasheva</author>
        <description><![CDATA[This paper systematically examines monitoring tools for containerized applications, microservices, and DevOps environments, and provides an experimental evaluation across two deployment scenarios. The study was conducted across two infrastructures: on-premises (Docker Swarm, Kubernetes) and cloud-based (Google Kubernetes Engine). A test application was deployed, and load and stress tests were performed using Apache JMeter and k6 with virtual user counts ranging from 100 to 1,000. Metrics including CPU, memory, I/O, and network utilization were collected using Prometheus, Grafana, cAdvisor, and Docker stats. Based on exploration and evaluation, 69 unique monitoring-related tools were identified and grouped into four non-mutually exclusive categories: container-based, cloud-based, microservices-oriented, and DevOps-focused (18, 25, 13, and 38 assignments, respectively; 94 assignments in total because some tools belong to multiple categories). A classification is presented according to visualization capabilities, supported metrics, and quality attributes, including performance, security, interoperability, and usability. Experimental results demonstrate that both monitoring stacks captured variations in processor utilization under load at the resolution configured by their default sampling intervals (15–30 s for Prometheus + cAdvisor and 60 s for Google Cloud Monitoring), with the 1,000-virtual-user stress bursts visible as sharp CPU peaks in both stacks. The contribution of this work is a practical guideline for selecting monitoring tools, developed on the basis of a reference experimental environment.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1936730</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1936730</link>
        <title><![CDATA[Software quality assurance in the era of Agentic AI: a systematic mapping study]]></title>
        <pubdate>2026-08-28T00:00:00Z</pubdate>
        <category>Systematic Review</category>
        <author>Randa Ouaarous</author><author>Imane Hilal</author><author>Abdellatif Mezrioui</author>
        <description><![CDATA[The growing complexity of software systems has increased demand for intelligent solutions in Software Quality Assurance (SQA). Agentic Artificial Intelligence (AI) offers autonomous decision-making, adaptive behavior, and proactive quality improvement, yet its applications across SQA remain fragmented. This systematic mapping study examines Agentic AI integration in SQA, analyzing application distribution, agent characteristics, and supported stakeholder roles to identify coverage, trends, benefits, limitations, and gaps. Following systematic mapping protocols and PRISMA, five digital libraries were searched, retrieving 37 primary studies. Data were classified through frequency analysis and qualitative categorization aligned with Agentic AI behavior, architecture, and autonomy dimensions. Applications concentrate in Product Assurance, mainly test generation and defect management, with fewer studies in Process or Planning activities. Most agents exhibit goal-oriented or diagnostic behaviors, moderate autonomy, and hybrid or multi-agent architectures. Benefits include efficiency, accuracy, and maintainability improvements; limitations involve scalability, generalizability, and transparency. Role coverage focuses on execution-level stakeholders, while managerial and leadership roles receive minimal support. Agentic AI shows potential for strengthening SQA intelligence and adaptability but remains unevenly distributed across activities and roles. Transparency, validation, and leadership support are priority research gaps for developing trustworthy, process-aware Agentic systems across the full SQA lifecycle.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1919566</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1919566</link>
        <title><![CDATA[An automated explainable decision-support system for multi-criteria assessment of functionalized implant coatings]]></title>
        <pubdate>2026-08-13T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Karlygash Alibekkyzy</author><author>Madina Batalova</author><author>Paweł Komada</author><author>Saule Belginova</author><author>Madina Bazarova</author><author>Talshyn Keribayeva</author><author>Indira Karymsakova</author><author>Yergazy Zheksenov</author>
        <description><![CDATA[IntroductionThe selection of functionalized implant coatings is a complex multi-criteria decision-making problem because biological, engineering, technological, and implementation-readiness indicators are heterogeneous and are frequently evaluated manually. This study proposes an automated and explainable computational decision-support framework for coating assessment and ranking.MethodsThe framework integrates data preprocessing, indicator normalization, weighted scoring, implementation-readiness assessment, sensitivity analysis, explainable classification, and visualization. Biological effectiveness, engineering and operational reliability, and manufacturability and scalability were used as the three evaluation domains. Published experimental data from 2015 to 2025 on titanium oxide nanostructures, hydrogel coatings, hybrid nanocomposites, and bioactive surface modifications were used for validation.ResultsThe framework generated reproducible domain scores, integral performance scores, implementation classes, and explainable rankings. TiO₂-Sr, Grad-Ti, and TiO₂-NT achieved the highest scores because of their balanced biological performance, engineering reliability, sterilization compatibility, and technological scalability. The ranking remained stable under variations in weighting factors.DiscussionThe proposed framework transforms heterogeneous biomedical data into a transparent and reproducible decision-support workflow. It can serve as the computational core of future biomedical material-selection platforms and support preliminary coating screening, technology-readiness evaluation, and data-driven engineering decisions.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1877426</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1877426</link>
        <title><![CDATA[AI-driven automation for CI/CD pipelines using attention-enabled reptile meta learning]]></title>
        <pubdate>2026-07-24T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Graeson Joshua Elijah</author><author>Mythily M</author><author>Fahmid Al Farid</author><author>Neeraj Addura G R</author><author>Salaja Silas</author><author>Elijah Blessing Rajsingh</author><author>Jia Uddin</author><author>Hezerul Bin Abdul Karim</author>
        <description><![CDATA[Healthcare systems increasingly rely on software to support critical operations such as patient monitoring, diagnostic decision-making, and real-time data management, making reliability, speed, and continuous availability essential. The software development and deployment phases, therefore, play a crucial role in ensuring that the healthcare applications remain stable and compliant with regulatory standards. However, traditional Continuous Integration/Continuous Deployment (CI/CD) pipelines are largely reactive, relying on manual testing, delayed feedback, and post-failure recovery mechanisms, which are time-consuming and prone to errors. Such limitations lead to deployment instability, inefficient resource utilization, and increased operational risk in mission-critical healthcare environments. To overcome these challenges, automated and intelligent approaches are required to enable proactive failure detection, faster recovery, and reliable deployment. In this regard, a novel AI-driven self-adaptive framework, Continuous Integration/Continuous Deployment Attention-enabled Reptile Meta-Learning (CICD-ARML), is proposed. The proposed system continuously learns from pipeline telemetry and adapts to dynamic deployment conditions, reducing dependency on manual intervention. Experimental results demonstrate that CICD-ARML achieves a 96.8% build success rate, reduces mean recovery time by 35%, reduces deployment risk by 59%, and achieves 94% prediction accuracy. These findings highlight the effectiveness of the proposed framework in enhancing the reliability, efficiency, and adaptability of healthcare software deployment systems.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1816972</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1816972</link>
        <title><![CDATA[An application of selective coding to create a theoretical framework on the factors that influence success or failure in adopting the SIG-UFRN software ecosystem]]></title>
        <pubdate>2026-07-02T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Ailson Medeiros Vasconcelos</author><author>Sandro Ronaldo Bezerra Oliveira</author>
        <description><![CDATA[IntroductionWe call Software Ecosystems (SECOs) the development of software around a common technology platform that results in a set of solutions or services for a consumer market. The Integrated Management System (SIG-UFRN) of the Federal University of Rio Grande do Norte (UFRN) is a technological platform that forms part of the UFRN SECO and is used by a considerable number of Federal Educational Institutions (FEIs) throughout Brazil. The scenarios of use of the technology platform involve a complex mix of technical, social, and business aspects that can lead to the adoption or rejection of the SIG-UFRN. If these aspects are not observed, they can harm the business activities of the FEI, waste resources, and result in penalties for public managers who make poor decisions. This research aimed to construct and validate a theoretical framework that represents the adoption of the SIG-UFRN based on technical, business, and social factors.MethodsTo achieve this objective, Selective Coding, as defined in the Grounded Theory (GT) method, was used, as well as Focus Groups for validation. Selective coding is the third stage of the GT. This refines and integrates concepts to construct a theoretical framework for a phenomenon. This coding process enabled the refinement and interrelation of the concepts inherent in the adoption scenarios of the SIG-UFRN. These concepts were identified in the previous open and axial coding stages of GT. Using techniques such as Storyline Memo, Core Category, and Paradigm Model, a theoretical framework could be constructed representing the adoption of SIG-UFRN.ResultsWe identified a theoretical framework that explained the adoption process of the SIG-UFRN, which was then validated with stakeholders.DiscussionThe results of this study could influence the success or failure of the adoption process and guide stakeholders in adopting SIG-UFRN best practices.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1811944</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1811944</link>
        <title><![CDATA[Vitality assurance in microservice architectures: introducing the health box testing methodology]]></title>
        <pubdate>2026-06-09T00:00:00Z</pubdate>
        <category>Original Research</category>
        <author>Yusuf Köse</author><author>Sin-Chun Ng</author>
        <description><![CDATA[IntroductionMicroservice architectures introduce complex testing challenges, particularly for grey failures — partial degradation scenarios in which infrastructure signals appear normal while user-facing journeys fail silently. Existing monitoring approaches lack the sensitivity to surface such failures early.MethodsWe introduce Health Box Testing (HBT), a runtime vitality assurance methodology that deploys synthetic Health Agents to execute representative user journeys against production-like environments. A proof-of-concept evaluation was conducted across 12 controlled grey-failure injection trials using Playwright-based agents in a dedicated test environment.ResultsHBT detected all 12 injected grey-failure trials (coverage 100%; median MTTD = 70 s, IQR 48.5-86.8 s). The baseline monitoring approach detected 7 of 12 trials (58.3%), with a conditional median MTTD of 397 s (IQR 344.0-404.5 s). A Wilcoxon signed-rank test on the 7 paired trials confirmed HBT was significantly faster (W = 0, p = 0.01562, two-sided), with a median paired reduction of 322 s.DiscussionHealth Box Testing provides a complementary, toolchain-agnostic framework for continuous vitality assurance that surfaces user-impacting failures earlier than conventional infrastructure monitoring, particularly for grey failures invisible to standard telemetry. Future work includes automated root-cause classification and integration with cognitive-load-aware alerting frameworks.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1814498</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1814498</link>
        <title><![CDATA[Platform engineering and internal developer portals: a multivocal literature review]]></title>
        <pubdate>2026-05-04T00:00:00Z</pubdate>
        <category>Systematic Review</category>
        <author>Mateen Ali Anjum</author>
        <description><![CDATA[Platform engineering has become the dominant approach to managing developer infrastructure at scale, with industry surveys indicating that 94% of organizations have adopted or plan to adopt dedicated platform teams. Despite this rapid practitioner uptake, academic research remains scarce: a systematic search across five major databases identified fewer than a dozen peer-reviewed papers from reputable venues that address platform engineering directly, while gray literature from foundations, vendors, and industry surveys is abundant. This study presents the first multivocal literature review (MLR) of platform engineering and internal developer portals, following published guidelines for including gray literature in software engineering reviews. The review synthesizes 88 sources across both peer-reviewed and gray literature, with sources explicitly tiered by provenance and gray literature assessed using the AACODS framework. Five research questions address the state of the literature, architectural components and patterns, success metrics and KPIs, adoption barriers, and the relationship between platform maturity and developer productivity. The synthesis yields a taxonomy of internal developer portal components grounded in 36 architecture-focused sources, an integrated metrics framework spanning DORA, SPACE, and developer experience dimensions, a comparative analysis of four platform engineering maturity models, and a quantification of the academic–practitioner divide: only 2 of 88 included sources (2.3%) originate from tier-1 venues with platform engineering as their primary topic, while practitioner communities have generated the authoritative definitions, frameworks, and measurement instruments as academic engagement lags by two to three years. A particularly striking gap concerns scorecards, the primary governance mechanism within IDPs, for which no peer-reviewed empirical evidence of effectiveness exists despite widespread commercial adoption. These findings carry implications for both researchers and practitioners. For the research community, the study identifies nine specific opportunities where empirical work is most needed, with a validated PE maturity model and a PE-specific measurement instrument representing the highest-impact contributions. For practitioners, the evidence supports treating platforms as products, combining delivery metrics with developer experience surveys, and designing golden paths as enablers instead of mandates.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1737008</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1737008</link>
        <title><![CDATA[The role of explainability throughout the MLOps lifecycle: review and research agenda]]></title>
        <pubdate>2026-05-01T00:00:00Z</pubdate>
        <category>Review</category>
        <author>Sule Tekkesinoglu</author><author>Matthias Wagner</author><author>Per Runeson</author>
        <description><![CDATA[As Machine Learning Operations (MLOps) adoption accelerates, systematic integration of explainability is imperative for reliability, transparency, and continuous quality assurance. This paper presents a scoping review examining how explainability is integrated across the MLOps lifecycle, encompassing data handling, model development, and deployment. Each phase is further analyzed through its subareas: data handling (data quality, data pre-processing, and data management), model development (training and pre-deployment auditing), and deployment (developer oversight and end-user interfacing). We identified several key touchpoints within each subarea where XAI methods address specific technical and operational challenges. The synthesis covers a wide range of topics, from explainable imputation and data filtering to fairness auditing in high-stakes decision-making. Findings reveal that although explainability is widely applied, it remains fragmented, with insufficiently validated reliability, and limited operationalization for regulatory compliance. Building on this analysis, we propose a research agenda for embedding continuous explainability throughout MLOps pipelines. Key directions include connecting explainability touchpoints across lifecycle phases, validating the reliability of XAI methods, and operationalizing explainability to meet regulatory requirements such as those defined in the EU AI Act. By framing explainability as an infrastructural mechanism for assurance rather than a post-hoc diagnostic feature, this work advances a lifecycle-spanning perspective on trustworthy and governable AI systems.]]></description>
      </item><item>
        <guid isPermaLink="true">https://www.frontiersin.org/articles/10.3389/fcomp.2026.1845840</guid>
        <link>https://www.frontiersin.org/articles/10.3389/fcomp.2026.1845840</link>
        <title><![CDATA[Editorial: Software specification and verification: models and tools]]></title>
        <pubdate>2026-04-22T00:00:00Z</pubdate>
        <category>Editorial</category>
        <author>Vincenzo Arceri</author><author>Nabendu Chaki</author><author>Agostino Cortesi</author><author>Novarun Deb</author>
        <description></description>
      </item>
      </channel>
    </rss>