DataFormats»Multi Net Format

Multi Net Format


Csv / fixed witdh


That proposal described a "free-format" specification that seems to me to probably be the easiest one to implement for both software developers and the people who use the software. When the delimiter is a comma, the format becomes a CSV (comma-separated-values, as can be produced and read by most spreadsheets) with a simple header at the beginning to provide variable names and any value labels that are present.

  1. It can easily handle large, complex, multivariate datasets in which there are many node attribute variables and highly multiplex relational networks.
  2. It's easy for people who use the software to create and edit data in this format because they can do most of the work with an ordinary spreadsheet like Excel or Quattrro.
  3. It's easy to work with large networks because it doesn't use adjacency matrices. Have you ever tried to count over to the 1237th column in an adjacency matrix? Okay, your networks aren't that big. How about the 268th column?
  4. It's easy to work with because it only describes the links that are present. If there is no link between node #472 and node #1237, you don't have to do anything about that link that doesn't exist -- it isn't necessary to put a zero in the 1237th column of the 472nd row of the adjacency matrix.
  5. It's easy to work with because it doesn't require a different matrix for each type of relationship.
  6. It's easy to work with because it doesn't make huge files filled mostly with zeros that indicate the absence of links. Since it only describes links that are present, the data files are a lot smaller than what would be required for adjacency matrices.

other info

 Multiple files?  

"..Bill Richards referenced, regarding some MultiNet data standards developed by a group at one of the Sunbelts ("