More details about this model
Thank you for your contribution! Could you please provide more details about this model, such as the description of the training dataset? What makes this model perform better than the one presented in the EMNLP2021 paper?
Thanks for your concern. As you know, the limitation of EMNLP2021 paper is the extracted sentenc is not always simile, the same problem is occurring in"Writing Polishment with Simile: Task, Dataset and A Neural Approach". Hence, we build our dataset using the model trained on the simile component(the tenor and vehicle ) dataset("Neural Multitask Learning for Simile Recognition"), which ensure the quality of our dataset.
Thanks for your reply. As far as I know, the dataset and the model from the "Neural Multitask Learning for Simile Recognition" is Chinese, while it appears that your uploaded model is suitable for English. Did you use translation to bridge the gap?
You're right, we built our dataset in Chinese, then we translated it into English using a professional translation tool.
Would it be possible to get access to your translated dataset? Appreciate any help you can offer with this.
Sorry, we can't offer you the dataset we used just yet. We have assembled a paper on some of the strategies for this task and have submitted it to ACL2023. However, our dataset is also based on "Writing Polishment with Simile: Task, Dataset and A Neural Approach".