Paper Reference Relevance Visualisation

Simply provide the title to any research paper to visualise all referenced papers scaled by relevance in an interactive network.

Each cited paper is assigned a score based on it's relevance to the root paper. This score determines the size of each node in the network.

The score is calculated with respect to the:

• Number of citations of a given paper in the root paper.
• Semantic similarity of the given and root paper's abstracts.
• Influence of a given paper to the root paper.

Semantic similarity is computed using Cohere's embedding function by computing the similarity between the embedded version of both abstracts.

Influence is a boolean variable determined by Semantic Scholar's isInfluential variable.

The colour of each node is determined by the intent of each paper given by the Semantic Scholar API.

• Background = Blue
• Method = Red
• Result = Green
• Root = Yellow
• None = Grey

Set environment variable for Cohere

Copy your Cohere API key from here.

Replace 'your_api_key' with copied key:

export COHERE_API_KEY='your_api_key'

Install requirements

pip install -r requirements.txt

Run main.py with your desried paper title

python main.py -t "PAPER TITLE HERE"

To just run with the stored JSON file

python main.py -d PATH_TO_FILE