mderank

This is code for paper: MDERank: A Masked Document Embedding Rank Approach for Unsupervised Keyphrase Extraction Data is from OpenNMT-kpg-release and SIFRank. (Inspec, DUC2001, SemEval2017 are from SIFRank).

Table of Contents

• Environment
• Usage
• Cite

Environment

Python 3.7
nltk 3.4.3
StanfordCoreNLP 3.9.1.1
torch 1.1.0
allennlp 0.8.4
pke 1.8.1
transformer 4.14.1
CUDA version 10.2

Usage

We use run.sh script to run MDERank.

sh run.sh

--checkpoint is the model used for predictions. Initial MDERank use bert-base-uncased.

Cite

If you use this code, please cite this paper

@article{DBLP:journals/corr/abs-2110-06651,
  author    = {Linhan Zhang and
               Qian Chen and
               Wen Wang and
               Chong Deng and
               Shiliang Zhang and
               Bing Li and
               Wei Wang and
               Xin Cao},
  title     = {MDERank: {A} Masked Document Embedding Rank Approach for Unsupervised
               Keyphrase Extraction},
  journal   = {CoRR},
  volume    = {abs/2110.06651},
  year      = {2021},
  url       = {https://arxiv.org/abs/2110.06651},
  eprinttype = {arXiv},
  eprint    = {2110.06651},
  timestamp = {Fri, 22 Oct 2021 13:33:09 +0200},
  biburl    = {https://dblp.org/rec/journals/corr/abs-2110-06651.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}