Detecting Knowledge Inconsistencies Across Text, Tables, and Knowledge Graphs

2026-07-28

AI Summary

This work addresses the problem of cross-modal knowledge inconsistency among text, tables, and knowledge graphs by proposing Kontrast, a novel framework that introduces the first systematic taxonomy of cross-modal knowledge conflicts and enables their automated detection and attribution. The approach integrates Text-to-SPARQL query generation, large language model reasoning, and knowledge graph alignment techniques to systematically compare table-based question answering results against knowledge graph evidence, thereby identifying distinct inconsistency types such as factual contradictions, structural omissions, and temporal mismatches. Extensive experiments across multiple table QA benchmarks demonstrate both the prevalence and informational value of such inconsistencies, establishing Kontrast as an effective tool and benchmark for knowledge auditing, complementation, and correction.

Abstract

Wikipedia and Wikidata are widely used for information access, LLM pre-training, and retrieval-augmented generation. Their knowledge is deeply connected but scattered across text, tables, and knowledge graphs. This raises a practical question: when these modalities disagree, how can we detect and explain the conflict? We study this problem as \emph{modality-level inconsistency detection}. We first introduce a taxonomy of cross-modal knowledge inconsistencies, covering information granularity differences, direct conflicts, temporal changes, and KG incompleteness. We then present \textsc{Kontrast}, an automatic framework that uses Text-to-SPARQL and LLM reasoning to compare table-based answers with KG evidence and categorize the resulting inconsistencies. Experiments on various Table-QA datasets show that cross-modal inconsistencies are common and informative. They reveal not only true knowledge conflicts, but also missing KG structure and temporal mismatches while being limited by Text-to-SPARQL errors and noise. Our analysis shows that text, tables, and KGs can complement and correct one another through systematic comparison. \textsc{Kontrast} provides a practical tool for large-scale knowledge auditing and establishes a benchmark for future work on cross-modal knowledge consistency. Code and data are available at https://github.com/ECLADATTA/KONTRAST.

Problem

• knowledge inconsistency
• cross-modal
• knowledge graph
• tables
• text

Innovation

• cross-modal inconsistency
• Text-to-SPARQL
• knowledge graph auditing
• LLM reasoning
• modality-level inconsistency detection