An Approach for Detecting AIGC Text Hallucinations Based on Knowledge Graph

This paper presents an approach for detecting AIGC(Artificial Intelligence Generated Content) text hallucinations based on a knowledge graph framework. Comparison algorithms are utilized to analyze text data and extract key features, which are mapped onto the knowledge graph for comprehensive analysis. By innovating the Jaccard Index formula, KGDH Jaccard Index formula is proposed to demonstrate the index of text hallucination generation. Through empirical evaluations and comparisons, the effectiveness and feasibility of this approach in identifying text hallucinations is demonstrated. This research contributes to the field of AI ethics and misinformation detection by providing a systematic framework for scrutinizing the credibility of AI-generated content.