On Creating an Interoperable Network Matrix



Complete self taught novice here.


Are there any popular standards, or commonly used formats for representing/saving/sharing/serializing a network?


I built a simple but very large (5.5m nodes, 16.7m edges) network matrix using my own python code. But if I wanted to use this matrix with other data science tools what format should I use?

Currently I have it in a giant csv file with headers and "1" representing edges. Which I think is called a binary matrix.


I want to make use of any benefits present in using binary files, or formatted flat files that would help me learn other tools.

Gabriel Fair

Posted 2017-12-05T18:11:37.113

Reputation: 197

No. There is no single universally popular input/output standard for networks. – Valentas – 2017-12-09T10:18:43.027

I'm trying to figure out if this could be one: http://tinkerpop.apache.org/

– Gabriel Fair – 2017-12-09T16:59:29.940



There is no single universally accept standard, however, many packages use edgelist for sparse (un)directed (un)weighted graphs. Regarding storing and sharing, I usually see compressed text files (in ascii format) instead of binary ones, due to binary compatibility problems.

Maybe you should take a look at this draft: Unravelling Graph-Exchange File Formats.


Posted 2017-12-05T18:11:37.113

Reputation: 21