Researchers have developed a new information-geometric framework to justify composite coherence metrics used in event-based narrative extraction. The study proposes a composite metric $C=\sqrt{A\cdot T}$, combining angular similarity of document embeddings ($A$) and topic proximity ($T$), and provides an axiomatic characterization for the geometric-mean combinator. Experiments across various corpora, embedding families, and topic models validate the framework, showing that the geometric mean is optimal and outperforms alternative methods. AI
IMPACT This research provides a theoretical foundation for improving narrative extraction, potentially enhancing how AI systems understand and generate coherent stories.
RANK_REASON The cluster contains an academic paper detailing a new framework and experimental results for narrative extraction.
Read on arXiv cs.IR (Information Retrieval) →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
