/Network-Code-Repository

This repository is created to maintain code repositories for the network science community.

GNU General Public License v3.0GPL-3.0

Network-Code-Repository

This is a centralized repository for network analysis tools, so to speak, the awesome list of network analysis recipes.

Created during the OpenNetSci Hackathon (May 25-26, 2019), with inputs from Nesreen K. Ahmed, Yong-Yeol Ahn, Tiago P. Peixoto, and Jean-Gabriel Young, we host a repository of metadata files for non-standard software. Our goal is to enhance the exposure of small tools and strengthen the network science community as an ecosystem.

With that, please feel welcomed to send pull requests with the metadata of your code to this repository!

Motivation

The vast majority of software developed by network scientists takes the form of small standalone code bases, distributed across researcher's homepages, hosting services, and code sharing site like GitLab, Bitbucket or GitHub. This practice encourages algorithm duplication, reduces reproducibility and favors a fragmentation of the research community. A good way of addressing these issues would be to create a large, searchable index of code bases, perhaps in the spirit of ICON or networkrepository.

Infrastructure

We maintain a list of JSON files that encode the metadata to each code repository. While there is no fancy frontend design yet, it is easy to bootstrap the JSON files contained in this repository into a beautiful-looking website with JavaScript and HTML, or simply generate an awesome plain list with a parser script. It is also easy to import the files into a database, such as MongoDB or PostgreSQL.

This repository is structured as follows.

  • library: contain code that is installable via a package manager, or importable from the language interface.
  • paper: contain code that is either a script, demo, or prototype directly from a paper.
  • standalone: contain code that can be compiled to a binary application, runnable on an operating system.
  • web: contain code to a web application that can be interacted through a browser.

Metadata

Each JSON file consists of the following fields:

  • name - String: Name of the repository.

  • desc - String: Short description of the repository.

  • category - String: Category of the repository; in the format of $category.$subcategory, connected via a dot. Within each category, the subcategories will be either general, visual, or special.

  • links - String[]: Links to the source code, homepage, or Git repository.

  • license - Object: Software license.

    • type - Type of license; typical open source licenses are MIT, Apache-2.0, or GPL-3.0. (SPDX short identifier is preferred) If you are not sure which one to choose, please visit: choose a license.
    • link - Link to description. Please paste the link to licenses that have been approved by the Open Source Initiative. For details, please visit: Licenses by Name.

  • languages - String[]: Programming languages supported, or based.

  • authors - Object[]: List of author of the repository. Each author item contains the following metadata.

    • name: Name of the author.
    • email: Email of the author.
    • site: Website of the author.
    • gscholar:

  • dates - Object

    • createdTime - Created time of the repository; be it a timestamp or just the calendar year.
    • lastUpdatedTime - Last updated time of the repository; be it a timestamp or just the calendar year.

  • io - Object: File I/O.

    • input: Input data formats that the tool accepts.
    • output: Output data formats that the tool generates.

  • community - Object: Features that associated with the scientific community.

    • tutorials: Link to tutorials that explain how this code work.
    • demos: Link to demos (e.g., visualizations) that explain how this code work.
    • pubs: Link to related papers that use this code.

  • tags - String[]: User-defined, subjective, tags; e.g., Python2.7, efficient, or hard-to-install, etc.

Future Work

  • Add more recipes to the repository!
  • Write a script that generates an awesome list out of these JSON files.
  • Build a command-line tool (similar to brew or apt-get) that can search repositories which match user-specified keywords.

Frontend

You can build your own website with your preferred features implemented. For example, Papers with code enables users to search code bases with keywords. They also show the GitHub stars that indicate a trending research work. Currently, the Network Science Education in Taiwan intiative has a searchable frontend for the indices listed in the repository.