ReTabAD: A Benchmark for Restoring Semantic Context in Tabular Anomaly Detection

2025-10-02

AI Summary

Existing tabular anomaly detection (AD) benchmarks lack semantic context—such as feature descriptions and domain knowledge—limiting model performance and interpretability. To address this, we introduce TabAD-Bench, the first AD benchmark explicitly supporting semantic context: it systematically integrates 20 annotated tabular datasets with structured textual metadata, including feature definitions, business rules, and domain constraints. We further propose TabAD-LLM, a zero-shot large language model (LLM) framework that performs context-aware anomaly detection without fine-tuning. TabAD-LLM jointly encodes tabular structure and semantic metadata to enable domain-informed reasoning. Extensive experiments demonstrate that incorporating semantic context significantly improves detection accuracy and explanation quality. This work establishes a standardized, semantics-enriched benchmark for tabular AD, provides a strong zero-shot baseline, and introduces a novel paradigm for context-driven anomaly detection.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

In tabular anomaly detection (AD), textual semantics often carry critical signals, as the definition of an anomaly is closely tied to domain-specific context. However, existing benchmarks provide only raw data points without semantic context, overlooking rich textual metadata such as feature descriptions and domain knowledge that experts rely on in practice. This limitation restricts research flexibility and prevents models from fully leveraging domain knowledge for detection. ReTabAD addresses this gap by restoring textual semantics to enable context-aware tabular AD research. We provide (1) 20 carefully curated tabular datasets enriched with structured textual metadata, together with implementations of state-of-the-art AD algorithms including classical, deep learning, and LLM-based approaches, and (2) a zero-shot LLM framework that leverages semantic context without task-specific training, establishing a strong baseline for future research. Furthermore, this work provides insights into the role and utility of textual metadata in AD through experiments and analysis. Results show that semantic context improves detection performance and enhances interpretability by supporting domain-aware reasoning. These findings establish ReTabAD as a benchmark for systematic exploration of context-aware AD.

Problem

• Restoring textual semantics in tabular anomaly detection
• Addressing lack of semantic context in existing AD benchmarks
• Enabling domain-aware reasoning through enriched metadata

Innovation

• Enriched datasets with structured textual metadata
• Zero-shot LLM framework leveraging semantic context
• Implementing classical, deep learning, and LLM algorithms