Dataset Validation Script

[ffl096]

We should provide a validation script that ensures datasets to be in the specified format. Non-exhaustive list of thinks to validate:

General structure

• Dataset must be a (optionally gzipped) plain text file.
• Dataset metadata (JSON object) on first line.
• Any node-related line must appear before edges.

Attribute-related

• Dates and datetimes (whenever a time attribute appears) should be in ISO 8601 format. The timezone must be specified.
• weight attributes must be a float or integer.

Suggestions

The validation script should issue improvement suggestions that are not necessarily errors but good practice to follow:

• A node is explicitly given, but does not has explicit attributes and appears inside an edge (i.e., can be inferred implicitly)