Prompt-based Logical Semantics Enhancement for Implicit Discourse Relation Recognition

The Code for the EMNLP 2023 main conference paper "Prompt-based Logical Semantics Enhancement for Implicit Discourse Relation Recognition".

Dependence Installation

pip install -r requirements.txt

Data

Pre-training data

We approximately collected 0.56 million explicit argument pairs from an unannotated corpus renowned as Gigaword. Please note the data is only utilized for academic research purposes.

Prompt-tuning data

We use PDTB 2.0 and CoNLL 2016 Shared Task to evaluate our models. If you have bought data from LDC, please put the PDTB data and CoNLL data in src/data/pdtb2 and /src/data/conll respectively.

Data Preprocessing

run the following commands, respectively:

unzip ./src/data/explicit_data/explicit_data.zip -d ./src/data/explicit_data

sh ./scripts/data_preprocess_pdtb2_4.sh

sh ./scripts/data_preprocess_pdtb2_11.sh

sh ./scripts/data_preprocess_conll.sh

Pre-training

run the following command:

sh ./scripts/pretrain_plse.sh

Prompt-tuning

For 4-way classification on PDTB 2.0, run the following command:

sh ./scripts/train_pdtb_4.sh

For 11-way classification on PDTB 2.0, run the following command:

sh ./scripts/train_pdtb_11.sh

For 14-way classification on CoNLL16, run the following command:

sh ./scripts/train_conll_14.sh

Bibliography

If you find this repo useful, please cite our paper.

@inproceedings{wang-etal-2023-prompt-based,
    title = "Prompt-based Logical Semantics Enhancement for Implicit Discourse Relation Recognition",
    author = "Wang, Chenxu  and
      Jian, Ping  and
      Huang, Mu",
    booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
    month = dec,
    year = "2023",
    address = "Singapore",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2023.emnlp-main.45",
    pages = "687--699",
}

lalalamdbf/PLSE_IDRR