ReFuGe: Feature Generation for Prediction Tasks on Relational Databases with LLM Agents

2026-01-25

AI Summary

This work addresses the challenge of insufficient feature information in relational database prediction tasks, where complex schema reasoning and vast search spaces often hinder effective feature engineering. To overcome this limitation, the authors propose ReFuGe, a novel framework that introduces a multi-agent large language model collaboration mechanism. ReFuGe employs three specialized agents—schema selector, feature generator, and feature filter—to iteratively generate and refine high-quality relational features in an unsupervised manner. By integrating dual filtering mechanisms based on both reasoning and validation, the approach eliminates the need for manual annotations while significantly enhancing feature effectiveness. Extensive experiments demonstrate that ReFuGe substantially improves predictive performance across multiple relational database benchmarks, underscoring its generality and practical utility.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Reasoning under Uncertainty: Relational Probabilistic Models
• Data Mining & Knowledge Management: Representing, Reasoning, and Using Provenance, Trust

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

Relational databases (RDBs) play a crucial role in many real-world web applications, supporting data management across multiple interconnected tables. Beyond typical retrieval-oriented tasks, prediction tasks on RDBs have recently gained attention. In this work, we address this problem by generating informative relational features that enhance predictive performance. However, generating such features is challenging: it requires reasoning over complex schemas and exploring a combinatorially large feature space, all without explicit supervision. To address these challenges, we propose ReFuGe, an agentic framework that leverages specialized large language model agents: (1) a schema selection agent identifies the tables and columns relevant to the task, (2) a feature generation agent produces diverse candidate features from the selected schema, and (3) a feature filtering agent evaluates and retains promising features through reasoning-based and validation-based filtering. It operates within an iterative feedback loop until performance converges. Experiments on RDB benchmarks demonstrate that ReFuGe substantially improves performance on various RDB prediction tasks. Our code and datasets are available at https://github.com/K-Kyungho/REFUGE.

Problem

• relational databases
• feature generation
• prediction tasks
• schema reasoning
• unsupervised feature learning

Innovation

• LLM agents
• relational feature generation
• schema reasoning
• unsupervised feature selection
• iterative feedback loop