https://github.com/declare-lab/kingdom

Domain Adaptation using External Knowledge for Sentiment Analysis
https://github.com/declare-lab/kingdom

adversarial-learning adversarial-networks deep-learning deep-neural-networks domain-adaptation opinion-mining sentiment-analysis

Last synced: about 4 hours ago
JSON representation

Domain Adaptation using External Knowledge for Sentiment Analysis

• Host: GitHub
• URL: https://github.com/declare-lab/kingdom
• Owner: declare-lab
• Created: 2020-05-02T00:12:02.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2023-07-06T21:50:04.000Z (over 1 year ago)
• Last Synced: 2024-04-16T04:16:01.568Z (7 months ago)
• Topics: adversarial-learning, adversarial-networks, deep-learning, deep-neural-networks, domain-adaptation, opinion-mining, sentiment-analysis
• Language: Python
• Homepage: https://arxiv.org/pdf/2005.00791.pdf
• Size: 31.9 MB
• Stars: 57
• Watchers: 7
• Forks: 15
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# KinGDOM: Knowledge-Guided DOMain adaptation for sentiment analysis (ACL 2020)

[_KinGDOM_](https://arxiv.org/abs/2005.00791.pdf) takes a novel perspective on the task of domain adaptation in sentiment analysis by exploring the role of external commonsense knowledge. It utilizes the ConceptNet knowledge graph to enrich the semantics of a document by providing both domain-specific and domain-general background concepts. These concepts are learned by training a graph convolutional autoencoder that leverages inter-domain concepts in a domain-invariant manner. Conditioning a popular domain-adversarial baseline method with these learned concepts helps improve its performance over state-of-the-art approaches, demonstrating the efficacy of the proposed framework.

![Alt text](KinGDOM.jpeg?raw=true "KinGDOM framework")

### Requirements

- scipy==1.3.1

- gensim==3.8.1

- torch==1.6.0

- numpy==1.18.2

- scikit_learn==0.22.2.post1

- torch_geometric==1.6.3

### Execution

Download ConceptNet filtered for English language from [here](https://drive.google.com/file/d/19klcp69OYEf29A_JrBphgkMVPQ9rXe1k/view?usp=sharing) and keep in this root directory.

Preprocess, train and extract graph features:

```bash

python preprocess_graph.py

python train_and_extract_graph_features.py

```

We provide pretrained graph features in the `graph_features` directory. Note that, executing the above commands will overwrite the provided feature files.

Train the main domain adaptation model:

```bash

python train.py

```

Some of the RGCN functionalities are adapted from https://github.com/JinheonBaek/RGCN

### Citation

Please cite the following paper if you find this code useful in your work.

```bash

KinGDOM: Knowledge-Guided DOMain adaptation for sentiment analysis. D. Ghosal, D. Hazarika, N. Majumder, A. Roy, S. Poria, R. Mihalcea. ACL 2020.

```