Chinese Metaphor Generation

Latest Version of Model and Data of our COLING paper are available Now!

Source code and dataset of

INLG 2022 oral: Nominal Metaphor Generation with Multitask Learning and

COLING 2022 oral: CM-Gen: A Neural Framework for Chinese Metaphor Generation with Explicit Context Modelling.

首个中文比喻数据集，可以用于中文比喻识别与中文比喻生成。

Chinese Metaphor Corpus

The first Chinese metaphor corpus serving both metaphor identification and generation. The definition of metaphor in Chinese is different with its English counterpart. So we cannot borrow metaphor resource in English and use it in Chinese metaphor processing.

We construct a big metaphor resoruce in Chinese with around 9000 metaphorical sentences with tenor and vehicle annotated. Some examples are shown below.

{"sent": "春风像一只彩笔，把整个世界勾勒得更加绚丽多彩", "tenor": "春风", "vehicle": "彩笔", "comparator": "像"}
{"sent": "秋天，不仅是一个丰收的季节，还让我们饱览了迷人的景色。秋天在秋姑娘的点缀后，像一颗发光的明珠！", "tenor": "秋天", "vehicle": "明珠", "comparator": "像"}
{"sent": "春风像个慈祥的母亲，拂着你的脸颊，使你感到舒畅，心旷神怡。", "tenor": "春风", "vehicle": "母亲", "comparator": "像"}
{"sent": "秋天，田里的稻谷黄澄澄的，像一片片金色的海洋。天空是蓝蓝的，像一片无边无际的大海。", "tenor": "天空", "vehicle": "大海", "comparator": "像"}

As shown here metaphor sentences and their components are identified. To enlarge the dataset, we merge the dataset published in Liu et al. 2018 with ours. If you use the dataset, please cite both our and their paper.

The dataset can be found in CMC/zh_CMC2.txt.

Label and sentence was splited by \t

Old version of CMC used in our INLG paper can be found here.

Chinese Literature Dataset

Chinese Literature corpus (CLC) is a corpus contains only chinese literatures. Literature only corpus is been needed in some NLP tasks, like processing some kinds of texts which are more frequnt viewed in literatures, or analyisis of the difference between literature and other texts (like mews and dialogues).

Chinese Literature Dataset consists of 7.31 millon sentences crawled from 1300+ chinese books, which has already been shuffled. Average length of all sentences is 29 tokens.

The download url is: https://pan.baidu.com/s/1XY72p_QHnYuP8-yQkOY0fA.

Model

Please refer to here.

citation

If you use our resource, do remember citing our paper:

@inproceedings{li-etal-2022-cm,
    title = "{CM}-Gen: A Neural Framework for {C}hinese Metaphor Generation with Explicit Context Modelling",
    author = "Li, Yucheng  and
      Lin, Chenghua  and
      Guerin, Frank",
    booktitle = "Proceedings of the 29th International Conference on Computational Linguistics",
    month = oct,
    year = "2022",
    address = "Gyeongju, Republic of Korea",
    publisher = "International Committee on Computational Linguistics",
    url = "https://aclanthology.org/2022.coling-1.563",
    pages = "6468--6479",
}

@misc{li-inlg-2022-nominal,
  doi = {10.48550/ARXIV.2206.05195},
  url = {https://arxiv.org/abs/2206.05195},
  author = {Li, Yucheng and Lin, Chenghua and Geurin, Frank},
  keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Nominal Metaphor Generation with Multitask Learning},
  publisher = {arXiv},
  year = {2022},
  copyright = {arXiv.org perpetual, non-exclusive license}
}