- Python 3.8
- AWS EC2
- Scihub Unofficial API
- CrossRef API
- Pull paywalled research papers
- Find additional citations
- Autosummarise abstract with GPT3
- Parse research findings with GPT3
I’m Not Reading All That!
We built our own Discord Bot to pull and parse information from research papers that are locked behind ridiculous paywalls using Scihub’s API directly to the Discord chat. We also connected the produced papers with CrossRef to also find all relevant metadata on it such as abstracts, which were then summarised using A.I. (GPT3 algorithm). This was built in Python within a few days using Discord.py (unofficial Discord wrapper for Python). See it in action on our Discord server.
1. How It Works
A Connection Of API’s
Our QOVES Discord Bot works by contacting the SciHub unofficial API to pull a research paper using its DOI. SciHub is an online repository of research papers gathered from paywalled publishers. After doing so, the bot queries the given DOI with CrossRef’s database to find all the relevant metadata it can on it, such as other papers that have cited this work, its abstract, author and year, all of which is data that we take and present nicely in the Discord chat. In later versions, a thumbnail of the paper was added which captures the first two
pages of the PDF sent back by SciHub, to only then paste them together and upload it back to an AWS S3 Bucket where it is hosted with a temporary URL. This bucket is wiped daily and acts as transient storage. In doing so, we bypass Discord’s embed rules where images can only be shown as embedded if it is hosted with a URL, rather than passing the image file back directly to Discord. Embedding the metadata, image and file links also results in nicer presentation in-chat.