Have way to see which packages need to be reviewed and enhanced

[JonoYang]

For example, if a package has unknown in the license expression, that package would be a candidate for metadata enhancement.

We need to determine if a package that has been mined and scanned is good as is or needs further review. There are two axis: depth and breadth. Breadth: do we have all the attribute we need (license, copyright, home, etc). Depth: do we have correct data (license clarity, license text and notice etc.)

This should be based on available code in SCTK and should feed a review TODO list.

[pombredanne]

See also the new --todo option in SCTK by @AyanSinhaMahapatra in aboutcode-org/scancode-toolkit#3346

[pombredanne]

@DennisClark ping.... We need some design on this

[DennisClark]

Consider a possible approach to the "breadth" concept:

A new field completion_level : a system-calculated value, from 0 to 99, that indicates how many of the critical attributes (license, copyright, home, etc) have data rather than null.

• the value should be calculated and set whenever a PurlDB entry is created or modified.
• the different critical attributes (to be determined and documented) would each have an assigned weight (for example, license=30, copyright=20, home=20, etc.) so that when an entry is well-defined the score would be 99.

Not sure how to approach "depth" yet, since that might be a challenge to automate, but it could be a similar score field, perhaps curation_level that is set by a process that validates the various critical fields. A zero value would indicate not curated yet, and therefore needs to be curated. Any other value, greater than zero, would be a result of a curation process, either manual or automated.