AnTenA: Actionable and Explainable Tensor Analysis System with Large Language Models

2026-06-26

AI Summary

This work addresses the challenge of interpretable pattern discovery in multidimensional tensor data when accurate labels or auxiliary metadata are unavailable. The authors propose AnTenA, a novel system that uniquely integrates large language models (LLMs) with tensor factorization to generate human-understandable explanations of latent patterns uncovered through co-clustering. By leveraging both task-agnostic and task-specific prompts, AnTenA enables forward and backward reasoning for hypothesis validation, eliminating the need for conventional metadata. The approach facilitates actionable and verifiable pattern discovery in fully unsupervised settings, demonstrating strong interpretability and effectiveness across diverse narrative-rich multidimensional datasets.

Abstract

Accurately explaining hidden patterns in multi-aspect data has typically been done by leveraging labels and/or accompanying auxiliary metadata. However, labels and auxiliary data may be inaccurate (e.g. nonstandard, inconsistent), insufficient (e.g. static tabular metadata for time-dependent recordings), or unavailable. % We propose \fullmethod (\method), which leverages the knowledge of large language models (LLMs) to explain the hidden patterns in human narratives. \method uses task-agnostic and task-specific prompts to explain extracted co-clustered latent patterns from tensor decomposition. To evaluate these explanations, we test the LLMs on forward and backward inference tasks. % Our demo system is available at https://github.com/dawonahn/ECML_PKDD_AnTenA.

Problem

• tensor analysis
• explainable AI
• large language models
• multi-aspect data
• pattern explanation

Innovation

• tensor decomposition
• large language models
• explainable AI
• co-clustering
• prompt engineering