BiGCN

Convolution over Hierarchical Syntactic and Lexical Graphs for Aspect Level Sentiment Analysis

Code and preprocessed dataset for EMNLP 2020 paper titled "Convolution over Hierarchical Syntactic and Lexical Graphs for Aspect Level Sentiment Analysis"

Requirements

Python 3.7
PyTorch 1.1.0
SpaCy 2.0.18
numpy 1.16.2

Usage

Download pretrained GloVe embeddings with this link and extract glove.840B.300d.txt into glove/.
Train with command, optional arguments could be found in train.py

python train.py --dataset rest16 --vocab_dir datasets/semeval16/rest16_ --save True

Citation

If you use the code in your paper, please kindly star this repo and cite our paper

@inproceedings{DBLP:conf/emnlp/ZhangQ20,
  author    = {Mi Zhang and
               Tieyun Qian},
  title     = {Convolution over Hierarchical Syntactic and Lexical Graphs for Aspect
               Level Sentiment Analysis},
  booktitle = {Proceedings of the 2020 Conference on Empirical Methods in Natural
               Language Processing, {EMNLP} 2020, Online, November 16-20, 2020},
  pages     = {3540--3549},
  year      = {2020},
  crossref  = {DBLP:conf/emnlp/2020-1},
  url       = {https://www.aclweb.org/anthology/2020.emnlp-main.286/},
  timestamp = {Thu, 19 Nov 2020 16:13:16 +0100},
  biburl    = {https://dblp.org/rec/conf/emnlp/ZhangQ20.bib},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

Note

Code of this repo heavily relies on ABSA-PyTorch and ASGCN

About

Mi Zhang and Tieyun Qian,Convolution over Hierarchical Level Sentiment Analysis,EMNLP2020

Languages

Language:Python 100.0%