This repository contains the code for the paper A Novel Perspective to Look At Attention: Bi-level Attention-based Explainable Topic Modeling for News Classification. The implementation is based on Pytorch and includes news classification and news recommendation tasks. We also provide a procedure to generate explainable topic for both news classification task and news recommender system.
Clone the repository and install the dependencies.
git clone git@github.com:Ruixinhua/ExplainedNRS.git
cd explainable_nrs
pip3 install -r requirements.txtDownload the MIND dataset from here. The dataset is licensed under the Microsoft Research License Terms. We suggest to put the dataset in the dataset folder as follows. Download the MIND dataset and GloVe embeddings manually if automatically download failed and put it in the dataset folder as follows.
mkdir dataset && cd dataset
# Download GloVe pre-trained word embedding
wget https://nlp.stanford.edu/data/glove.840B.300d.zip
sudo apt install unzip
unzip glove.840B.300d.zip -d glove
rm glove.840B.300d.zip
cd ../
python3 experiment/preprocess/download_mind.py --mind_type=small --data_dir=dataset/MIND- Dataset by MIcrosoft News Dataset (MIND), see https://msnews.github.io/.
- Reference news recommendation repository by yusanshi, see https://github.com/yusanshi/news-recommendation