Many OData primitive types are JSON Schema primitive types combined with a set of restrictions, e.g. Edm.Int64 is

• "type":"integer",
• "minimum":-9223372036854775808,
• "maximum":9223372036854775807
  or Edm.Guid is
• "type":"string",
• "pattern":"^[0-9a-fA-F] {8}

  -[0-9a-fA-F]

  {4}-[0-9a-fA-F]{4}

  -[0-9a-fA-F]

  {4}

  -[0-9a-fA-F]

  {12}

  $"

There are basically three options:
1) place these in definitions in edm.json and reference them there
2) inline these definitions at each occurrence of the OData primitive type
3) just combine the JSON Schema primitive type with an OData-specific format, e.g.

• type: string
• format: guid
  or
• type: integer
• format: int64

The third approach is the most readable and in sync with what Swagger does, see http://swagger.io/specification, section Data Types, and we can try to get these formats into the next version of JSON Schema, see https://github.com/json-schema/json-schema/wiki/%22format%22-suggestions