Of Human Criteria and Automatic Metrics: A Benchmark of the Evaluation of Story Generation
arXiv:2208.11646
Abstract
Research on Automatic Story Generation (ASG) relies heavily on human and automatic evaluation. However, there is no consensus on which human evaluation criteria to use, and no analysis of how well automatic criteria correlate with them. In this paper, we propose to re-evaluate ASG evaluation. We introduce a set of 6 orthogonal and comprehensive human criteria, carefully motivated by the social sciences literature. We also present HANNA, an annotated dataset of 1,056 stories produced by 10 different ASG systems. HANNA allows us to quantitatively evaluate the correlations of 72 automatic metrics with human criteria. Our analysis highlights the weaknesses of current metrics for ASG and allows us to formulate practical recommendations for ASG evaluation.
43 pages, 38 figures. Proceedings of the 29th International Conference on Computational Linguistics (COLING 2022)
References in corpus (7)
- CTRL: A Conditional Transformer Language Model for Controllable Generation
- BARTScore: Evaluating Generated Text as Text Generation
- Transformer-based Conditional Variational Autoencoder for Controllable Story Generation
- Narrative Interpolation for Generating and Understanding Stories
- What are the best systems? New perspectives on NLP Benchmarking
- From the Token to the Review: A Hierarchical Multimodal approach to Opinion Mining
- Beam Search with Bidirectional Strategies for Neural Response Generation