PID Kernel Information Profile Management WG: AI applications

You are here

28 Nov 2019
Group(s) submitting the application: 
Meeting objectives: 

PID Kernel Information (PID KI) caches a small amount of information at a PID resolver making this information highly accessible to third parties and thus contribution to the practical implementation of FAIR data principles.  The information is defined through a profile. In this session the chairs have chosen to focus on Artificial Intelligence (AI) as a broad and emerging discipline-independent application area to illuminate use cases and their underlying data discovery, integration, and use workflows.  These workflows drawn from AI will be used to inform necessary (and unnecessary) profile differentiation. The chairs encourage engagement by AI experts in this session.

Meeting agenda: 

 

  1. Introduction and current status (10 min)

  2. Use cases from AI: overview (30 min.)

    1. Use cases gathered before the session

    2. Additional use cases 

  3. Role of PID KI in AI workflows for data discovery, integration, and use (50 min.)

  4. Planning next steps (10 min.)

Target Audience: 

The topic at hand is of interest to a variety of stakeholders in the areas of data generation, curation and usage. The group’s outputs can be of use to scientific infrastructure providers, client tool builders and scientific users. We particularly encourage feedback from practitioners in AI for use in any domain.

Brief introduction describing the activities and scope of the group: 

 

PID Kernel Information is a limited set of information (metadata) about a data object to which a PID refers. This associated information resides within the PID resolution system, that is, is separate from any domain-specific metadata that often resides at the repository where the data object resides. Through numerous discussions, the PID Kernel Information WG has converged on the underlying architectural principles for PID Kernel Information, building on the power of the combined RDA Recommendations on PID Information Types and Data Type Registries.

The PID Kernel Information WG produced a recommendation that contains i) guiding principles for identifying information appropriate as PID Kernel Information, ii) an exemplar PID Kernel Information profile, iii) several use cases, and iv) architectural considerations. PID Kernel Information is defined as the set of attributes stored within a PID record. It supports smart programmatic decisions that can be accomplished through inspection of the PID record alone.

The WG has now entered a second phase to further define the mechanisms by which to define, manage and make proper use of profiles of Kernel Information. The second phase was kicked off at RDA P14 with the intention to provide a follow-on recommendation specifically focused on profile management, with AI applications as broad motivational drivers. A PID Kernel Information profile normatively defines the metadata elements that are in Kernel Information records specific to a user community or application and thus go beyond the generic profile that emerged from the work of the first group phase.

Short Group Status: 

The group has been established for 6 months now and seeks openly further input from new use case providers, in particular.

Type of Meeting: 
Working meeting
Remote participation availability: 
Yes