DataFormats»Process

Process

Some proposals on how to create standards

(none of which seem to be exclusive. Each should have its own page)

INSNA take role, form committe

  • Pres of INSNA, created a blue ribbon committee to discuss data standards for network data, and to write a "white paper" with proposals. Many of these discussions here on SOCNET have been a great start in that direction.
    • We need to ensure that the committee has at least a couple good software engineers (good = significant industry experience) to make sure that what we'll be working out is (a) doable and (b) "future-proof"
    • At the same time, we need to bear in mind that there is a big difference between the needs of scientists, who write programs in order to get their real work done, and software engineers, whose real work is writing programs. Software engineers often adopt technically sophisticated approaches that are appropriate for writing programs or designing formats that will only ever be read by other software engineers. When considering formats for the network community, it is important to bear in mind that they have to be accessible to people who mostly don't develop software for a living. This means that sometimes a less powerful but more transparent approach is needed. If we develop a more powerful approach that requires a lot of study to understand, it is unlikely to gain widespread adoption.

Email discussion

  • Start listserv for discussing standard and formats

Collect and organize existing knowlege

  • Make a wiki to collect and organize existing knowledge

Tag libraries to help catagorize and search

  • library
  • tool
  • platforms
  • open-source
  • language: {R, Java, C/C++, Matlab, ...}
  • capability: {visualization, analysis} [maybe some more refined tags if

necessary, e.g. {zoom/pan, blockmodeling, centrality} ]

  • license: {GPL, LGLP, BSD, pay, etc.}
  • original authors (credits)
  • maintainers / listserv to contact for help
  • file examples (for formats)
  • code / usage examples (for formats)

Define common format and translation parsers

settle on a file format (e.g., GraphML) and then write a program or set of programs that convert between that format and other common ones, such as Pajek format, ASCII matrix formats, and others

  • It would allow people to publish data in any format: if you don't like the format, change it into GraphML (or whatever) and from there into your favorite format.
  • It would allow people to use other people's software with their data.
  • It would allow people to use other people's data with their software.
  • It would allow people to move data from one piece of software to another.
  • And it would, in effect, establish a de facto standard for network data in the form of the chosen format, which with luck the wider community might then adopt even if it doesn't receive an official endorsement from INSNA.

Build online service for translation / validation

  • online web service with an HTTP interface. We would even save ourselves the trouble of implementing the same conversion library in different programming languages. If there is an agreement on the interface of the service, different people can implement it different ways and host it at various places around the world.

Format Camp / Workshop

  • one day workshop/brainstorming session on the topic of network data standards.
  • hold it in a way that we could also include participation from people not able to travel via a wiki, live chat, perhaps even a Skype group call/stream from the conference.
  • suggest holding this one day workshop in an Open Space / "Camp" format - with no specific set agenda or speakers, rather as a highly

collaborative and open-ended workshop. We'll meet, have introductions, then break up into smaller groups around specific topics, documenting the discussions via public tools (wikis for example) and then coming back together as a group.