BERT-E2E-ABSA

Exploiting BERT End-to-End Aspect-Based Sentiment Analysis

Requirements

• python 3.7.3

• pytorch 1.2.0

• transformers 2.0.0

• numpy 1.16.4

• tensorboardX 1.9

• tqdm 4.32.1

• some codes are borrowed from allennlp (https://github.com/allenai/allennlp, an awesome open-source NLP toolkit) and transformers (https://github.com/huggingface/transformers, formerly known as pytorch-pretrained-bert or pytorch-transformers)

Architecture

• Pre-trained embedding layer: BERT-Base-Uncased (12-layer, 768-hidden, 12-heads, 110M parameters)

• Task-specific layer:

• Linear

• Recurrent Neural Networks (GRU)

• Self-Attention Networks (SAN, TFM)

• Conditional Random Fields (CRF)

Dataset

• Restaurant: retaurant reviews from SemEval 2014 (task 4), SemEval 2015 (task 12) and SemEval 2016 (task 5)

• Laptop: laptop reviews from SemEval 2014

Quick Start

• Reproduce the results on Restaurant and Laptop dataset:

  # train the model with 5 different seed numbers
  python fast_run.py

• Train the model on other ABSA dataset:

1. place data files in the directory ./data/[YOUR_DATASET_NAME].

2. set TASK_NAME in train.sh as [YOUR_DATASET_NAME].

3. train the model: sh train.sh

• (** New feature **) Perform pure inference/direct transfer over test/unseen data using the trained ABSA model:

1. place data file in the directory ./data/[YOUR_EVAL_DATASET_NAME].

2. set TASK_NAME in work.sh as [YOUR_EVAL_DATASET_NAME]

3. set ABSA_HOME in work.sh as [HOME_DIRECTORY_OF_YOUR_ABSA_MODEL]

4. run: sh work.sh

Environment

• OS: REHL Server 6.4 (Santiago)

• GPU: NVIDIA GTX 1080 ti

• CUDA: 10.0

• cuDNN: v7.6.1

Citation

If the code is used in your research, please star our repo and cite our paper as follows:

@inproceedings{li-etal-2019-exploiting,
    title = "Exploiting {BERT} for End-to-End Aspect-based Sentiment Analysis",
    author = "Li, Xin  and
      Bing, Lidong  and
      Zhang, Wenxuan  and
      Lam, Wai",
    booktitle = "Proceedings of the 5th Workshop on Noisy User-generated Text (W-NUT 2019)",
    year = "2019",
    url = "https://www.aclweb.org/anthology/D19-5505",
    pages = "34--41"
}