This repo is the slightly modified version of https://github.com/karpathy/arxiv-sanity-preserver.

Crawling and Organizing arxiv papers

1. Build virtual env.

conda env create -f arxiv-env.yml

2. Set your search keywords(list) at utils.py. (crawler will search for articles that contain the keywords in the abstract.)

class Config(object):
    search_list = ["graph convolution","graph neural network"]
    # search query : [graph] AND [convolution] OR [graph] AND [neural] AND [network]
    save_pdf_by_title = True
    # if save_pdf_by_title = False, the pdf file will be saved as 1805.07857v2.pdf
    save_pdf_by_months = True
    # if save_pdf_by_months = False, the files will not be organized by months. (will be organized by years.)

3. Run source crawling.sh

conda activate arxiv-env

python fetch_papers.py --max-index=1000  # maximum number of papers
python download_pdfs.py

# repeat
python fetch_papers.py --max-index=1000
python download_pdfs.py

conda deactivate

4. (Optional) Put papers on Mendeley library.

You can write simple review at note and organize papers by using tags. (searching papers based on notes and tags)