Skip to main content

Introduction

The Base Product Specification (BPS) provides a generic, interoperable framework for representing products of all kinds — whether Data Products, AI Products, software components, services, or physical goods.

This specification establishes a conceptual foundation upon which domain-specific product descriptors (e.g., DPDS/DPROD for Data Products, APDS/APROD for AI Products) may be consistently defined and integrated.

Purpose​

The purpose of the BPS is to:

  • Provide a neutral ontology for describing products across domains.
  • Support consistent governance and lifecycle management.
  • Enable the development of unified product marketplaces that integrate heterogeneous product types.
  • Facilitate interoperability between emerging and existing specifications.

Scope​

The specification is intended for use in:

  • Enterprise data and AI platforms that must govern and expose both Data and AI Products.
  • Software and services ecosystems where consistent product semantics reduce fragmentation.
  • Cross-industry collaborations requiring a standard product description independent of implementation or domain.

The BPS does not prescribe technical deployment details for any specific domain; instead, it defines an abstract foundation from which such details may be derived.

Motivation​

Despite the ubiquity of products in human enterprise, existing specifications are domain-specific and fragmented:

  • Data Products are described by DPDS (deployment) and DPROD (semantic description).
  • AI artifacts are partially described by ONNX, PMML, Model Cards, or risk frameworks such as NIST AI RMF.
  • Physical products are covered by ontologies such as GoodRelations or ETIM.
  • Service and software products use diverse cataloging approaches.

This fragmentation hinders governance, discoverability, and interoperability.
The BPS addresses this challenge by defining a base layer applicable to all product types.

Position in Standards Landscape​

The BPS is intended to complement existing domain specifications rather than replace them.
It provides:

  • Alignment with W3C vocabularies (DCAT, PROV-O) for data and provenance.
  • Compatibility with ISO/IEC AI and software engineering standards.
  • Extensibility to support new product categories such as digital twins or simulation models.

Expected Audience​

The specification is targeted at:

  • Standards bodies developing domain-specific product descriptors.
  • Enterprises building holistic governance frameworks for Data and AI.
  • Researchers and practitioners in data management, AI ethics, and product lifecycle management.
  • Architects and engineers designing unified product marketplaces.

Scope and Limitations​

The Base Product Specification (BPS) is a meta-framework for describing products.
It defines the universal split between:

  • PDS (Product Descriptor Specification): deployment blueprint
  • PROD (Product Ontology/Description): semantic blueprint

This split is applicable across all product domains.

What BPS is for​

  • Providing a conceptual skeleton for any future product specification.
  • Ensuring that all product types can be described in a consistent and interoperable manner.
  • Serving as a foundation layer for domain-specific specifications.
  • Aligning diverse product ecosystems (data, AI, software, physical) under a common meta-model.

What BPS is not for​

  • BPS is not a domain-specific implementation spec.
    • It does not define dataset schemas (data).
    • It does not define bias metrics or model cards (AI).
    • It does not define electrical tolerances or logistics (physical).
    • It does not define API endpoints or manifests (software).
  • BPS cannot be used on its own to describe products in practice.
  • BPS does not replace domain specifications such as DPDS/DPROD (Data Products) or future AIPDS/AIPROD (AI Products).

Strategic Implication​

BPS should be viewed as the grammar of product descriptors.

  • Grammar alone cannot produce meaning.
  • When combined with a domain vocabulary (e.g., DPDS/DPROD for Data, AIPDS/AIPROD for AI), it enables precise and actionable specifications.

Therefore, any implementation MUST extend BPS through domain-specific profiles before use in production environments.

In summary:
The BPS introduces a universal product kernel — a minimal yet extensible set of concepts that can be specialized into domain-specific descriptors. It aims to become a foundation for coherence across the fragmented landscape of product specifications.