error taxonomy design

The root cause analysis (RCA) community suffers from a critical shortage of large-scale, open-source, multimodal benchmark datasets, hindering rigorous method evaluation and advancement. To address this, we introduce LEMMA-RCA—the first cross-domain, multimodal, open-source RCA dataset tailored for IT/OT systems, encompassing realistic failure scenarios from microservices, water supply, and wastewater treatment. It comprises hundreds of system entities and fine-grained causal relationship annotations. Uniquely integrating four heterogeneous modalities—time-series metrics, logs, topology graphs, and alerts—it supports both offline/online and unimodal/multimodal RCA evaluation. Built via distributed monitoring, multi-source alignment, controllable fault injection, and causal graph annotation, the dataset ensures high fidelity and representativeness. Extensive evaluation across eight baseline methods demonstrates that multimodal joint modeling improves average F1-score by 23.6%. LEMMA-RCA is publicly released, establishing a new community benchmark for RCA research.

This work addresses the challenges of fusing heterogeneous modalities—such as time-series metrics and textual logs—and the inherent difficulty large language models face in processing continuous temporal data for root cause analysis in cloud infrastructure failures. To this end, the authors propose a multimodal diagnostic framework that aligns time-series performance indicators with the embedding space of pretrained language models through temporal semantic compression, a gated cross-attention alignment encoder, and a retrieval-augmented generation mechanism. This integration enables automated root cause localization informed by historical knowledge. Experimental evaluation across six cloud system benchmarks demonstrates that the proposed method achieves a diagnosis accuracy of 48.75%, significantly outperforming existing approaches, particularly in complex, multi-fault scenarios.

This work addresses the limited interpretability of existing fault diagnosis models, which often fail to explain why a fault occurs or how to rectify it, and the challenge of directly applying large language models due to the semantic gap between industrial time-series signals and natural language. To bridge this gap, the authors propose S2S-FDD, a novel framework featuring a signal-to-semantic transformation operator that maps temporal characteristics—such as trends, periodicity, and anomalies—into natural language descriptions. Integrated with a multi-round tree-based diagnostic mechanism, the framework fuses historical maintenance records and dynamic signal queries to enable zero-shot, interpretable fault diagnosis. The approach supports human-in-the-loop feedback for iterative refinement and demonstrates strong effectiveness and practicality in zero-shot settings on multiphase flow industrial data.

To address the challenge of manual diagnosis for anomalous trace faults in microservice systems, this paper proposes TraFaultDia—a novel framework that establishes, for the first time, a multi-system, cross-domain few-shot anomaly trace classification paradigm. TraFaultDia integrates Model-Agnostic Meta-Learning (MAML), Graph Neural Network (GNN)-based trace representation learning, multi-task few-shot classification, and a cross-system fault pattern alignment mechanism. It achieves high-accuracy fault type identification on unseen systems using only ten labeled samples per class. Extensive experiments on TrainTicket and OnlineBoutique demonstrate that TraFaultDia attains an average accuracy of 93.26% (92.19% on novel tasks) in intra-system settings and 85.20% (84.77% on novel tasks) in cross-system transfer scenarios—significantly enabling zero-effort localization of faulty components and root causes.

To address the insufficient fusion of multimodal monitoring data (logs, metrics, traces) and poor task adaptability in microservice systems—leading to inaccurate root cause localization and fault type identification—this paper proposes a task-oriented multimodal fault diagnosis framework. Our method introduces: (1) a task-oriented learning mechanism that dynamically weights modality-specific representations according to diagnostic subtasks (e.g., instance localization vs. fault type classification), thereby enhancing modality advantages on demand; and (2) cross-modal contrastive learning coupled with graph-level data augmentation, which mitigates label scarcity by randomly masking normal instances. Evaluated on two real-world datasets, our framework achieves a 55.94% improvement in Hit Rate@1 and a >4.08% gain in F1-score over state-of-the-art approaches, demonstrating superior diagnostic accuracy and robustness.

This study addresses the challenge in predictive maintenance of complex systems, where heterogeneity and redundancy among monitoring variables often obscure fault signals and undermine model interpretability. The authors propose a semantic feature segmentation framework that leverages domain knowledge to decompose variables into canonical components—carrying essential predictive signals—and residual components containing marginal information. These components are further grouped semantically according to functional mechanisms such as throughput, latency, and pressure. This approach introduces, for the first time, a domain-driven decomposition of the feature space that preserves semantic meaning while emphasizing fault-relevant features. Experimental results demonstrate that canonical components consistently outperform residual ones across diverse temporal configurations, achieving prediction accuracy comparable to models using all features or PCA-based representations, while exhibiting superior structural cohesion and operational interpretability.

Intermittent failures in continuous integration (CI) pipelines are notoriously difficult to diagnose, leading to wasted resources and reduced development efficiency. This work proposes FlaXifyer, a few-shot learning approach that integrates the interpretable AI technique LogSift to fine-tune pretrained language models on pipeline logs using only 12 labeled examples per failure class. The method simultaneously predicts failure categories and pinpoints critical log entries indicative of root causes. Evaluated on 2,458 real-world CI failures, FlaXifyer achieves a Macro F1 score of 84.3% and a Top-2 accuracy of 92.0%, reducing the required log inspection effort by 74.4%. Furthermore, it successfully identifies the underlying fault in 87% of cases, demonstrating its effectiveness in accelerating failure diagnosis with minimal labeled data.

This study addresses the challenge faced by operators in industrial settings who struggle to rapidly locate relevant troubleshooting procedures from vast volumes of technical documentation matching specific fault symptoms. To tackle this issue, the work proposes a retrieval-augmented generation (RAG)-based conversational assistance system, which is validated for the first time in a large-scale maritime cyber-physical system to demonstrate RAG’s practical efficacy in complex fault diagnosis scenarios. Experimental results show that the proposed approach significantly improves both the speed and accuracy of operator responses. Furthermore, the study underscores the necessity of incorporating cross-validation mechanisms to ensure the reliability of AI-generated recommendations, thereby offering actionable guidelines for deploying trustworthy AI-assisted decision-making in high-risk industrial environments.

Program analyses often lack robustness in the face of code changes. This work introduces, for the first time, a unified framework grounded in category theory that formalizes programs and their properties as categorical objects, capturing various forms of robustness—such as variable renaming and semantic refinement—via structure-preserving functors. Two implementation pathways are proposed: one lifts constructions from restricted computational models to general-purpose programs, while the other ensures stability in the composition of robust operators within algebraic program analyses. The framework not only uncovers common principles underlying loop summarization and termination analysis but also provides a theoretical foundation and predictability guarantees for developing program analyses that are more resilient to program transformations.

This study addresses the challenge of effectively monitoring early-stage agent systems, where structural flaws often obscure task-level errors. The authors propose a three-dimensional (quality, suitability, efficiency) and three-granularity (intra-run, inter-run, structural) monitoring and triaging framework tailored for low-maturity agent systems. They introduce a novel system maturity staging model based on the coefficient of variation and monitoring granularity, integrated with a severity classification adapted from FMEA to guide human review. The resulting transferable monitoring architecture supports document-driven, multi-stage workflows, enhanced by a synthetic testbed with controlled error injection. Experimental results demonstrate that structural defects significantly mask task-level signals; 97% of issues can be automatically traced, with only 2% requiring human intervention, and each granularity level precisely identifies its corresponding defect type (coefficients of variation: 0.02, 1.25, and 0.00, respectively).