visualisation_authors

Visualisation on co-authorship on pulications

• Size of nodes: Number of papers published (per author)
• Size of edges: Number of papers published together (pairwise)
• Color: Institution, legend disabled because it occupied nearly the whole screen (a lot of institutions)
• Details: simple histogram (publications per year)

Techonogies used

• Front-End: CSS, html, javscript (echarts, bootstrap)
• Middleware: NodeJS (Express, PythonShell)
• Back-End: Python (numpy, pandas, lecache, ...)
• Embedding
  • Embedding is a low-dimensional representation of high-dimensional data.
  • All embedding techniques attempt to reduce the dimensions of data, but meanwhile to preserve the "key" information in the data.
  • In this use case, the attempt is to obtain coordinates which can be presented on a 2-dimension graph, via using embedding techniques through a high-dimension adjacency matrix.

Dataflow

• Data pre-processing - get the proper data format
• Scrape author's ID from Semantic Scholar
• Transform processed data to adjacency matrix for further embedding, to get coordinates
• Plot embedded graph and statistic data

How to run the website

Installation with anaconda

The installation with conda is quite easy. This should

conda env create -f environment.yml
conda activate authors

# Install the required packages
pip install -r ./api/requirements.txt

Manual installation

Installing the requirements:

# Install nodejs can also be done with binary if you do not have sudo rights
sudo apt-get install nodejs
sudo apt-get install npm

# Install the required packages
pip install -r ./api/requirements.txt

npm install

Open the server localy:

# Run server
nodejs main.js