Know Your Intent: An Autonomous Multi-Perspective LLM Agent Framework for DeFi User Transaction Intent Mining

2025-11-19

AI Summary

Identifying DeFi transaction intents remains highly challenging due to smart contract complexity, heterogeneity between on-chain and off-chain data, and opaque transaction logs—leading to insufficient semantic understanding in existing approaches. To address this, we propose the first multi-agent large language model (LLM)-based framework for DeFi intent understanding. Our method comprises three core components: (1) a fine-grained, grounded-theory-driven intent taxonomy; (2) a dynamic meta-planner that orchestrates domain-specialist agents for task decomposition and multi-perspective collaborative analysis; and (3) a cognitive evaluation mechanism to mitigate hallucination and ensure verifiable, auditable reasoning. Experimental results demonstrate that our framework significantly outperforms conventional machine learning models, monolithic LLMs, and single-agent baselines in both accuracy and interpretability. This work establishes a methodological breakthrough for DeFi intent recognition, advancing robust, explainable, and scalable intent inference in decentralized finance.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Multiagent Learning
• Natural Language Processing: (Large) Language Models

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: Architectures and workflows that use LLMs for crowd work
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

As Decentralized Finance (DeFi) develops, understanding user intent behind DeFi transactions is crucial yet challenging due to complex smart contract interactions, multifaceted on-/off-chain factors, and opaque hex logs. Existing methods lack deep semantic insight. To address this, we propose the Transaction Intent Mining (TIM) framework. TIM leverages a DeFi intent taxonomy built on grounded theory and a multi-agent Large Language Model (LLM) system to robustly infer user intents. A Meta-Level Planner dynamically coordinates domain experts to decompose multiple perspective-specific intent analyses into solvable subtasks. Question Solvers handle the tasks with multi-modal on/off-chain data. While a Cognitive Evaluator mitigates LLM hallucinations and ensures verifiability. Experiments show that TIM significantly outperforms machine learning models, single LLMs, and single Agent baselines. We also analyze core challenges in intent inference. This work helps provide a more reliable understanding of user motivations in DeFi, offering context-aware explanations for complex blockchain activity.

Problem

• Mining user transaction intents in DeFi from complex smart contracts
• Overcoming opaque hex logs and multifaceted on/off-chain factors
• Addressing limitations in semantic insight from existing methods

Innovation

• Multi-agent LLM system for DeFi intent analysis
• Dynamic task decomposition with Meta-Level Planner
• Cognitive Evaluator mitigates hallucinations for verifiability