GRACE: LLM-Grounded Semantic Metric Spaces for Scalable Mixed-Data Clustering

2026-08-07

AI Summary

This work addresses the challenge of clustering mixed tabular data, where semantic heterogeneity between continuous and discrete features hinders traditional methods from capturing unobserved yet conceptually valid associations. To overcome this limitation, the authors propose a one-shot large language model (LLM) semantic grounding mechanism that leverages multi-view attribute-value queries to retrieve external knowledge, thereby mapping heterogeneous features into a unified semantic metric space. This external semantic representation is cross-validated against internal statistical evidence from the data. Crucially, the approach decouples costly LLM invocations from iterative optimization, ensuring computational scalability while significantly improving both clustering accuracy and conceptual interpretability. Empirical evaluation demonstrates that the method outperforms 11 established baselines, achieving state-of-the-art performance.

Abstract

Clustering mixed tabular data requires a unified metric space to bridge the inherent heterogeneity between continuous numerical measurements and discrete categorical symbols. Traditionally, algorithms rely entirely on dataset-internal statistics to estimate categorical relationships, which confines the learned metric to empirical co-occurrences and ignores conceptually obvious yet statistically unobserved affinities. Although LLMs offer external world knowledge, applying their text-centric reasoning to highly abstract tabular concepts presents significant challenges. Bridging this modality gap to construct a semantically complete metric typically requires embedding LLMs into iterative metric learning loops to dynamically optimize cross-modality representations. This incurs intractable computational overhead, forcing a compromise between semantic enrichment and scalability. Therefore, we propose GRACE, an LLM-grounded framework for scalable mixed-data clustering. GRACE shifts semantic acquisition to the attribute-value level via a multi-perspective LLM querying strategy, mapping heterogeneous values into knowledge-informed descriptions. Crucially, this one-shot grounding extracts general-purpose semantic representations that embed heterogeneous attributes into a unified space, decoupling expensive LLM invocation from iterative optimization. Furthermore, GRACE cross-validates these external semantics against dataset-internal statistical evidence to ensure alignment with the dataset-specific cluster structure. Ultimately, GRACE matches the scalability of conventional statistics-driven baselines while achieving superior clustering accuracy and conceptual interpretability over 11 competing methods. The source code is available at https://github.com/develop-yang/GRACE-GRACE-A

Problem

• mixed-data clustering
• semantic metric space
• LLM-grounded representation
• tabular data
• scalability

Innovation

• LLM-grounded semantics
• mixed-data clustering
• unified metric space
• one-shot semantic grounding
• scalable clustering