AcademicNexus aims to build an intelligent academic search engine leveraging GPT-3.5-turbo. Our initial focus is on Arxiv, a popular repository for academic papers, due to its accessible API and easy-to-parse search results format.
- Keyword extraction: GPT-3.5-turbo processes the title and abstract of the user-provided paper and extracts relevant keywords.
- Arxiv search: The extracted keywords are used to search for related papers on Arxiv.
- Paper ranking: GPT-3.5-turbo ranks the search results based on their titles and relevance to the user's query.
- Summary generation: The top-ranked papers' abstracts are processed, and GPT-3.5-turbo generates a summary highlighting the relationships between the user's query paper and the related papers.
- Create virtual environment.
conda create -n academic_nexus python=3.10
- Clone the repository and install dependency.
git clone git@github.com:prismleong/AcademicNexus.git
cd AcademicNexus
pip install -r requirements.txt
- Paste your openai api key in main.py
# fill in your openai api key
openai.api_key = "sk-..."
- Run the application:
streamlit run main.py
- paste the arxiv id and click search!