💻 Category

cs

2787 papers

💻 computer science

BM25 and Dense Retrieval Are Complementary for Portuguese Clinical Text: An Empirical Study of Hybrid RAG Across 500 Clinical Queries

This empirical study demonstrates that for Portuguese clinical decision support, hybrid retrieval combining BM25 and dense methods significantly outperforms single-strategy approaches by leveraging their complementary strengths, while also validating automated LLM-based evaluation and deterministic citation verification to ensure accuracy and reduce hallucinations.

Igor Eduardo2026-08-21
💻 computer science

Brazilian-PHI: Benchmarking Checksum-Validated Recognizers for CPF, CRM, CNS, CNPJ, RG, CEP, and Phone in Portuguese Clinical Text

This paper introduces Brazilian-PHI, the first benchmark for detecting seven types of Brazilian personal health information in clinical text, demonstrating that custom Presidio recognizers with checksum validation significantly outperform default tools and large language models in accuracy, latency, and compliance with LGPD requirements.

Igor Eduardo2026-08-21
💻 computer science

Deconstructing the Trolley Problem: From Static Triage to a Two-Tier Hierarchical Evaluation Framework for Constructive Moral Intelligence

This paper proposes a Two-Tier Hierarchical Evaluation Framework that transcends the static trade-offs of the classic trolley problem by integrating a non-compensatory geometric optimization tier for proactive physical interventions at t=-1 with a smooth transition to a consequentialist fail-safe tier at t=0, thereby reframing machine ethics from passive bystander calculation to active co-design of moral risk management.

Shinya Iida2026-08-21
💻 computer science

National Scale Disaster Response Optimization Engine Using Advanced Data Structures

This paper presents the National Scale Disaster Response Optimization Engine (NSDR-OE), a system leveraging an ensemble of eight advanced data structures to achieve real-time spatial indexing, urgency prioritization, and resource scheduling with O(log n) complexity, demonstrating a 231× speedup over linear baselines and sub-200 ms latency in both synthetic and live seismic event scenarios.

Vikas Maral, Kavya Bhand, Kabir Khanuja, Pranav Rana2026-08-21
💻 computer science

Structural Leakage in Host Intrusion Alert Corpora: An Evaluation Framework for ATT&CK Technique and Tactic Mapping

This paper introduces a leakage-aware evaluation framework for mapping Wazuh host intrusion alerts to MITRE ATT&CK tactics and techniques, demonstrating that standard evaluation methods overestimate performance due to structural data leakage from rule identifiers and alert templates, while showing that tactic-level prediction remains robust when these biases are controlled.

Emad Sherif2026-08-21
💻 computer science

Explainable AI-Based Labor Planning Framework for Distributed Transportation and Fulfillment Networks

This study proposes and validates an explainable AI-based framework that integrates demand forecasting, productivity analysis, and multi-objective optimization to unify labor planning decisions across distributed transportation networks, demonstrating its ability to reduce staffing shortfalls and unnecessary labor hours while providing transparent, manager-readable recommendations.

Nakshi Das2026-08-21
💻 computer science

Adversarial Evaluation of a Two-Layer Anonymization Pipeline Against Record-Linkage Attacks

This paper empirically evaluates the security of a two-layer anonymization pipeline that combines syntactic privacy constraints for record-level data with differential privacy for aggregate queries against realistic record-linkage attacks, demonstrating that the absence of a joint formal guarantee necessitates direct adversarial assessment across diverse datasets and knowledge scenarios.

Mohammed Sayim Khalil2026-08-21
💻 computer science

Applied Mathematical Robustness Analysis of Maximum-Likelihood Pairwise Ranking for Comparison-Driven Intelligent Systems

This paper investigates the robustness of maximum-likelihood pairwise ranking estimators against coordinated, budget-constrained perturbations using the Adaptive Subset Selection Attack (ASSA) heuristic, revealing that ranking fragility is highly data-dependent and regime-sensitive rather than universally predictable.

Junyi Yao, Zihao Zheng, Jiayu Long2026-08-20