Neuro-Symbolic Evaluation of Text-to-Video Models using Formal Verification

arXiv 2024 33.8 benchmark, method

TLDR

Introduces NeuS-V, a neuro-symbolic metric using formal verification to evaluate text-to-video alignment, outperforming existing metrics by 5x correlation with human judgments.

Reasoning

The paper addresses an important gap in text-to-video evaluation by focusing on temporal fidelity and alignment via formal verification, and provides a dataset and human-correlation results. However, its scope is limited to evaluation metrics rather than proposing or improving generative models, and the abstract does not demonstrate broader applicability beyond alignment checking.

Read-first score

Read-first score 33.8, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 0.

Recency 8%

75.1

Uses a gentle age decay so recent papers surface without erasing older foundations. 2024

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=benchmark,dataset,evaluation,metric

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,dataset,github

Topical relevance 42%

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Field roles

Candidate

Rank sensitivity

Stability: volatile; rank range: 384.

Keyword Scores

world model

world simulator

generative world model

interactive world model

video world model

world dynamics prediction

model-based reinforcement learning world model

Resources

Paper