Ask HN: What's a decent, human-friendly file format for property graph data?

1 points by rambojazz ↗ HN
If there is one single thing that I envy "semantic web" people for, is Turtle. Turtle is very human friendly, simple to read and to edit manually. It supports prefixes. It supports multiple graphs (TriG). It supports inline types/modifiers, for example "hello"@en and "10"^^<http://www.w3.org/2001/XMLSchema#integer>. It supports literal vs reference links, for example :Anna :age "20"; :knows :Bob (first is a literal, second is a link to another node). It has comments. It is standard and pretty much every triplestore can parse it.

People using property graphs don't seem to have anything like this. CSV, JSON, YAML? What do you guys use? I would like to use Turtle with property graphs but the usage of URI is just awful and perhaps only 1 or 2 proprietary (non free) property graphs can read it.

I'm curious about what you property-graph people use for serialization.

2 comments

[ 4.7 ms ] story [ 14.4 ms ] thread
Thank you for commenting. I've only ever tried CSV, JSON really. I like the first one, pg-format, the only problem I see is that all the labels and properties must be on a single line. What if I have a lot of properties, or very long strings, that I want to split into multiple lines for ease of manual editing? Also, does it take strings for IDs or only integers?