Hippasus: Effective and Efficient Automatic Feature Augmentation for Machine Learning Tasks on Relational Data

TL;DR

Hippasus tackles the core challenge of enriching a base relational table with informative features from related tables by decoupling path exploration, join execution, and feature selection. It combines lightweight statistical signals with Large Language Model (LLM) semantic reasoning to prune join paths before materialization and uses suffix-Yannakakis-based multi-way joins to efficiently consolidate features. The Feature Description Generator and hybrid statistical–semantic feature selection enable both meaningful and predictive feature augmentation, leading to substantial downstream performance gains (up to 26.8% over baselines) while maintaining competitive runtimes. The approach demonstrates strong practicality for large schemas and diverse domains, with flexible LLM choices and robust ablation showing the value of semantic context, statistical grounding, and efficient join processing.

Abstract

Machine learning models depend critically on feature quality, yet useful features are often scattered across multiple relational tables. Feature augmentation enriches a base table by discovering and integrating features from related tables through join operations. However, scaling this process to complex schemas with many tables and multi-hop paths remains challenging. Feature augmentation must address three core tasks: identify promising join paths that connect the base table to candidate tables, execute these joins to materialize augmented data, and select the most informative features from the results. Existing approaches face a fundamental tradeoff between effectiveness and efficiency: achieving high accuracy requires exploring many candidate paths, but exhaustive exploration is computationally prohibitive. Some methods compromise by considering only immediate neighbors, limiting their effectiveness, while others employ neural models that require expensive training data and introduce scalability limitations. We present Hippasus, a modular framework that achieves both goals through three key contributions. First, we combine lightweight statistical signals with semantic reasoning from Large Language Models to prune unpromising join paths before execution, focusing computational resources on high-quality candidates. Second, we employ optimized multi-way join algorithms and consolidate features from multiple paths, substantially reducing execution time. Third, we integrate LLM-based semantic understanding with statistical measures to select features that are both semantically meaningful and empirically predictive. Our experimental evaluation on publicly available datasets shows that Hippasus substantially improves feature augmentation accuracy by up to 26.8% over state-of-the-art baselines while also offering high runtime performance.

Figures (7)

• Figure 1: Overview of Hippasus.
• Figure 2: Prompt template for Feature Description Generator.
• Figure 3: Prompt template for Path Explorer (semantic table scoring).
• Figure 4: Prompt template for Feature Selector.
• Figure 5: Performance comparison of Suffix-Yannakakis approach. (a) Speedup over all paths across different datasets. (b) Speedup over binary join by path length on the Northwind dataset.