AI agent: Jacky
Author: Jackson Wu (jackson911007@gmail.com )
Date: Jan. 14 2026

• Packages require:
  (1) camelot
  (2) pdfplumber
  (3) google.generativeai
  (4) arxiv
  

Here we built a automated literature review pipeline:

• There are two main scripies in the "Code" folder web_scrapping_arxiv.py and Jacky.py
  
• web_scrapping_arxiv.py uses "arxiv" package developed by Cornell University Library (https://pypi.org/project/arxiv/) to do crawling from arxiv database
  
• Jacky.py is the code to do pdf information extraction using Google Gemini
  <Please first visit "Gemini api studio" to create a api key>
  

Code explanation:

• web_scrapping_arxiv.py did web crawling by defining prompt to search on arvix website. It then downloads all related literatures and stored in user-defined folder
  
• Jacky separates to two main parts:
  (1) Read in pdf file --> sparse the text and extract the text from table ---> combine all the table and contect information
  (2) Apply Gemini to digest the text message. To do so, you need to define clear message (prompt) to Gemini and design the output
  
• Jacky outputs a csv file that collects the information extracted from pdf based on user-defined prompt