Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/stuartpb/w5-graph

World Wide Web Who's Who
https://github.com/stuartpb/w5-graph

Last synced: 7 days ago
JSON representation

World Wide Web Who's Who

Awesome Lists containing this project

README 

        
# w5-graph



World Wide Web Who's Who



## Structure



The "nodes" directory contains YAML files describing nodes, with the name of

the node the file contains data for (with spaces replaced with underscores)

followed by ".yaml". (A node can specify its name with a "name" property in

the file if its name doesn't map cleanly to a filename: in this case, the

filename is merely a best-effort attempt to describe the node's name, for

human consumers.)



Properties on nodes:



### type



At the high level, node types basically break down into `person` nodes, and

all other nodes.



The other types of nodes have looser types that can be vaguely broken down into

two different kinds:



- Subjects

  - `project`

  - `spec`, `standard` (`standard` is basically a body of `spec` nodes)

  - `browser`

  - `library`

- Organization units

  - `location` (physical location)

  - `company`, `organization` (closed or semi-closed membership)

    - If you work for it, it's a company, if you're a member of it, it's an

      organization

  - `network` (open membership)

    - Twitter, Tumblr, and GitHub are networks

  - `forum` (closer-knit community)

  - `issues` (bug tracker)

  - `site` (static domain)



### wikidata



The identifier for the node's subject on [Wikidata][], if available.



[Wikidata]: https://www.wikidata.org/



### info



A short bio or description of the node's subject.



### image



The URL of an image that can represent that node. (Should work when cropped to

a circle.)



### href



A canonical URL for the node's subject.



### urls



A map of other URLs related to the node, by subject (ie. "Twitter" for a

person's Twitter account, "crbug" for a specification's implementation

issue on the Chromium issue tracker).



### rels



Edges from this node to others, as a map of names or arrays of names. See the

guidelines below for how to structure rels.



## Guidelines



These are just some rules of thumb to follow when defining nodes to keep the

graph meaningful.



### Try to copy the closest applicable existing convention



There should only really be a divergence from the existing nomenclature when

the divergent property could be used in several other cases to better

understand the meaning in the graph on the aggregate (eg. showing all

companies)



### One rel per pair, directed by action



Directed edges generally go from smaller nodes to bigger ones, keeping the

number of edges out from each node roughly on the same order of magnitude. If

a `rels` property is a very large array, it's probably the wrong way around.



Edges come from the point of concern: it's a bigger deal for a spec to be

"implemented by" a browser than a browser to "implement" a spec (although

*people* **do** implement specs). Similarly, "forked from" goes from the fork

to the source.



These concerns are usually active rather than passive: "requires" is a rel,

"is required by" is just "requires" backwards.



### Companies are not their products, and products are not their networks



"GitHub, Inc." is different from GitHub the website, and working *using* GitHub

is different from working *to develop* GitHub.



Since the goal of this is to visualize connections, companies have distinct

nodes separate from their flagship products, and those products are only listed

if they have a significant role in the communication and development activities

of the people on the graph (so Google-the-search-engine doesn't have a node).



For situations where the set of people working on a product is a significant

subset of emplyees within the company, they should be grouped based on what

*component* of the project they work on (eg. the frontend or backend).