Notes from the BOF session at DLF

These are also taken by Leah.

Moderators: Abrams, Stanescu

Lou Reich: one possible use case for GDFR is to view the related representation scheme.

Reich: It would be interesting to know whether there is a loss in the transform from one format to another - is this a relationship? Ans: Yes, if someone fills it in.

Abrams: We will have to define a topology of services, such as a watermark
service, at a coarse level. This will not be a service registry - you'll need a different data model and search model for this.

Comment: assign the identifier in GDFR so it works with a service registry.

Comment: data quality will be an issue for the service area in the registry. You may need as many as 10 people to review data quality. Ans: If people fill in service references, can they validate that the sources that they used actually work?

Expressing the relationship at an appropriate level of granularity will later allow machine actionable services.

Question: Is there something here that talks about a community? Most assessments are based within a community. Answer: we may need this.

Discussion related to what will be allowable as an assessment.

Evan Owens: "Defect assessment will be a growing industry" and knowing what to do with a defect will be interesting. Abrams: instead of calling it a defect - call it a variant to take away the subjective part of it. You can discuss how this affects accessibility. We can treat bugs as features.

Reich: Are you going to define the critical features? If we define the semantic units it will be too much trouble.

Another class of service is "I know how to fix this problem" - that would be a service definition.

Consider how this relates to Jhove.