LLM-Assisted Logic Rule Learning: Scaling Human Expertise for Time Series Anomaly Detection

2026-01-27

AI Summary

Traditional unsupervised time-series anomaly detection methods often fail to align with the nuanced demands of supply chain operations, while manual expert analysis does not scale to millions of products. This work proposes an interpretable anomaly detection framework powered by large language models (LLMs), which systematically encodes domain expertise into deterministic symbolic rules through three stages: LLM-guided knowledge annotation, iterative generation and refinement of logical rules, and business-category-aware enhancement. To our knowledge, this is the first approach to enable large-scale deployment of LLM-driven interpretable rules. The method outperforms conventional unsupervised techniques in both accuracy and interpretability, and compared to end-to-end LLM-based detection, it offers significantly lower latency, reduced computational cost, and greater deployment stability—making it well-suited for large-scale production environments.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Time series anomaly detection is critical for supply chain management to take proactive operations, but faces challenges: classical unsupervised anomaly detection based on exploiting data patterns often yields results misaligned with business requirements and domain knowledge, while manual expert analysis cannot scale to millions of products in the supply chain. We propose a framework that leverages large language models (LLMs) to systematically encode human expertise into interpretable, logic-based rules for detecting anomaly patterns in supply chain time series data. Our approach operates in three stages: 1) LLM-based labeling of training data instructed by domain knowledge, 2) automated generation and iterative improvements of symbolic rules through LLM-driven optimization, and 3) rule augmentation with business-relevant anomaly categories supported by LLMs to enhance interpretability. The experiment results showcase that our approach outperforms the unsupervised learning methods in both detection accuracy and interpretability. Furthermore, compared to direct LLM deployment for time series anomaly detection, our approach provides consistent, deterministic results with low computational latency and cost, making it ideal for production deployment. The proposed framework thus demonstrates how LLMs can bridge the gap between scalable automation and expert-driven decision-making in operational settings.

Problem

• time series anomaly detection
• supply chain management
• domain knowledge
• scalability
• unsupervised anomaly detection

Innovation

• LLM-assisted rule learning
• logic-based anomaly detection
• time series interpretability
• expert knowledge encoding
• scalable anomaly detection