README.md

Backend Documentation

The backend is written using Django. Papers are stored in a PostgreSQL database. Search performance is increased using Elasticsearch. Setup is simplified using Docker.

File Structure

only important files are included

backend
│   README.md
│   manage.py
│   docker-compose.yml  # development compose file
│   Dockerfile  # development dockerfile
|   docker-entrypoint.sh  # entrypoint used to wait for db and perform migrations
│
└───backend
│   |
│   |   settings.py  # backend configuration
│   
└───data
|   |
│   |   <empty, but put your jsonl files here>
│
└───graphgenerator
    |
    |   documents.py  # documents used for elasticsearch index
    |   models.py  # models used for Django ORM
    |   relevance.py  # relevance functions for graph element selection
    |   serializers.py  # serializers for models
    |   similarities.py  # similarity functions for graph similarity calculation
    |   urls.py  # url maps for api
    |   views.py  # api functions for searching/graph/lookup/bulk lookup 
    |
    └───authors  # author search urls/views
    |   
    └───management  # custom django commands for database population
    |
    └───migrations  # database migrations

Configuration

For configuration, use backend/backend/settings.py. See Django settings docs for detailed information.

Graph customization

If you only wish to customize the graph, the only two relevant files are graphgenerator/relevance.py and graphgenerator/similarities.py. When generating a graph, both of them have the following responsibilities.

Relevance - Given a source node, decide which papers should be shown.

Similarity - Given a list of papers, calculate their pairwise similarities.

Details on how to implement custom relevance/similarity-functions can be found in the respective files.