Extracting Events Like Code: A Multi-Agent Programming Framework for Zero-Shot Event Extraction

2025-11-17

AI Summary

Zero-shot event extraction (ZSEE) faces critical challenges including structural incompleteness and schema violations. To address these, we propose the “Event-as-Code” paradigm, which compiles event schemas into executable class definitions, and design a multi-agent collaborative framework that decouples the process into four sequential, iterative phases: retrieval, planning, code generation, and validation—enabling deterministic structural verification and refinement. Leveraging large language models (LLMs), our approach integrates schema-aware prompting, dynamic feedback loops, and a dedicated validation agent to ensure high-fidelity, structured extraction. Extensive experiments across five domains and six state-of-the-art LLMs demonstrate that our method significantly outperforms existing zero-shot baselines, achieving superior performance in trigger identification, argument filling, and schema consistency—yielding outputs that are more complete, accurate, and strictly compliant with target event schemas.

Technology Category

• Natural Language Processing: Information Extraction
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• 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

Zero-shot event extraction (ZSEE) remains a significant challenge for large language models (LLMs) due to the need for complex reasoning and domain-specific understanding. Direct prompting often yields incomplete or structurally invalid outputs--such as misclassified triggers, missing arguments, and schema violations. To address these limitations, we present Agent-Event-Coder (AEC), a novel multi-agent framework that treats event extraction like software engineering: as a structured, iterative code-generation process. AEC decomposes ZSEE into specialized subtasks--retrieval, planning, coding, and verification--each handled by a dedicated LLM agent. Event schemas are represented as executable class definitions, enabling deterministic validation and precise feedback via a verification agent. This programming-inspired approach allows for systematic disambiguation and schema enforcement through iterative refinement. By leveraging collaborative agent workflows, AEC enables LLMs to produce precise, complete, and schema-consistent extractions in zero-shot settings. Experiments across five diverse domains and six LLMs demonstrate that AEC consistently outperforms prior zero-shot baselines, showcasing the power of treating event extraction like code generation. The code and data are released on https://github.com/UESTC-GQJ/Agent-Event-Coder.

Problem

• Zero-shot event extraction faces incomplete outputs from direct LLM prompting
• Existing methods produce misclassified triggers and schema violations
• Complex reasoning requires structured approach for domain-specific understanding

Innovation

• Multi-agent framework decomposes event extraction tasks
• Event schemas represented as executable class definitions
• Iterative refinement through collaborative agent workflows