mini-arxiv

Usage:

uvicorn main:app --reload --port=8123 [debug]
uvicorn main:app --port=8123 [prod]

# to get response
curl --location --request GET 'http://127.0.0.1:8123/papers' \
--header 'Content-Type: text/plain' \
--data-raw '{
    "n_return": 5
}'

curl --location --request GET 'http://127.0.0.1:8123/papermatch' \
--header 'Content-Type: text/plain' \
--data-raw '{
    "user_string": "BERT",
    "n_return": 5
}'

arxiv-sanity for NBX Dashboard.

Arxiv-Sanity can show recent papers, this is for NBX dashboard and implments the following apis:

/papers

API for getting latest papers

/papermatch

Performs tf-ids + fuzzy NN matching to get papers from the user query.

[WIP] get_papers_by_authors(author, n_return = 10)

Returns a json-able dictionary for the author searched.

Args:
    author (str): string shared by the user
    n_return (int): number of latest papers to return.

/categorymatch

Returns a json-able dictionary for papers in a certain category.

NOTE: The input will be the complete category string as given in the ./taxonomy file. Eg. send {"user_string": "Artificial Intelligence"} and not cs.AI. Provide the category as an option to the user.

Files

There are two files:

• main.py: The file that has the above mentioned APIs
• prepare.py: prepares the dataset, expected to be run as a CRON job sometime post midnight PST, that's when the papers hit arxiv servers.