Unsupervised Explanation Generation via Correct Instantiations


  • Sijie Cheng Shanghai Artificial Intelligence Laboratory Fudan University
  • Zhiyong Wu Shanghai Artificial Intelligence Laboratory
  • Jiangjie Chen Fudan University
  • Zhixing Li Full Truck Alliance
  • Yang Liu Institute for AI Industry Research, Tsinghua University Department of Computer Science and Technology, Tsinghua University
  • Lingpeng Kong Shanghai Artificial Intelligence Laboratory The University of Hong Kong




SNLP: Interpretability & Analysis of NLP Models


While large pre-trained language models (PLM) have shown their great skills at solving discriminative tasks, a significant gap remains when compared with humans for explanation-related tasks. Among them, explaining the reason why a statement is wrong (e.g., against commonsense) is incredibly challenging. The major difficulty is finding the conflict point, where the statement contradicts our real world. This paper proposes Neon, a two-phrase, unsupervised explanation generation framework. Neon first generates corrected instantiations of the statement (phase I), then uses them to prompt large PLMs to find the conflict point and complete the explanation (phase II). We conduct extensive experiments on two standard explanation benchmarks, i.e., ComVE and e-SNLI. According to both automatic and human evaluations, Neon outperforms baselines, even for those with human-annotated instantiations. In addition to explaining a negative prediction, we further demonstrate that Neon remains effective when generalizing to different scenarios. The resources of Neon are available at: https://github.com/Shark-NLP/Neon.




How to Cite

Cheng, S., Wu, Z., Chen, J., Li, Z., Liu, Y., & Kong, L. (2023). Unsupervised Explanation Generation via Correct Instantiations. Proceedings of the AAAI Conference on Artificial Intelligence, 37(11), 12700-12708. https://doi.org/10.1609/aaai.v37i11.26494



AAAI Technical Track on Speech & Natural Language Processing