A Generative Approach for Script Event Prediction via Contrastive Fine-Tuning


  • Fangqi Zhu Harbin Institute of Technology, Shenzhen
  • Jun Gao Harbin Institute of Technology, Shenzhen
  • Changlong Yu Independent Researcher
  • Wei Wang Independent Researcher
  • Chen Xu Beijing University of Technology
  • Xin Mu Peng Cheng Laboratory
  • Min Yang Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences
  • Ruifeng Xu Harbin Institute of Technology, Shenzhen Peng Cheng Laboratory Guangdong Provincial Key Laboratory of Novel Security Intelligence Technologies




SNLP: Sentence-Level Semantics and Textual Inference, SNLP: Applications


Script event prediction aims to predict the subsequent event given the context. This requires the capability to infer the correlations between events. Recent works have attempted to improve event correlation reasoning by using pretrained language models and incorporating external knowledge (e.g., discourse relations). Though promising results have been achieved, some challenges still remain. First, the pretrained language models adopted by current works ignore event-level knowledge, resulting in an inability to capture the correlations between events well. Second, modeling correlations between events with discourse relations is limited because it can only capture explicit correlations between events with discourse markers, and cannot capture many implicit correlations. To this end, we propose a novel generative approach for this task, in which a pretrained language model is fine-tuned with an event-centric pretraining objective and predicts the next event within a generative paradigm. Specifically, we first introduce a novel event-level blank infilling strategy as the learning objective to inject event-level knowledge into the pretrained language model, and then design a likelihood-based contrastive loss for fine-tuning the generative model. Instead of using an additional prediction layer, we perform prediction by using sequence likelihoods generated by the generative model. Our approach models correlations between events in a soft way without any external knowledge. The likelihood-based prediction eliminates the need to use additional networks to make predictions and is somewhat interpretable since it scores each word in the event. Experimental results on the multi-choice narrative cloze (MCNC) task demonstrate that our approach achieves better results than other state-of-the-art baselines. Our code will be available at https://github.com/zhufq00/mcnc.




How to Cite

Zhu, F., Gao, J., Yu, C., Wang, W., Xu, C., Mu, X., Yang, M., & Xu, R. (2023). A Generative Approach for Script Event Prediction via Contrastive Fine-Tuning. Proceedings of the AAAI Conference on Artificial Intelligence, 37(11), 14056-14064. https://doi.org/10.1609/aaai.v37i11.26645



AAAI Technical Track on Speech & Natural Language Processing