The volume of data sharing is dramatically increasing in both personal and business contexts. Often, multiple disparate parties are addressing the same underlying raw data through different applications; permissioning models; use restrictions; and with different ultimate objectives.
Raw data is increasingly being stored by third parties (in one or multiple locations in the cloud or in company systems) rather than under the direct control of its ultimate owners.
As a result, explicit management of metadata (“information about resources”) has become increasingly important, not only for discovery and control, but also for correctness, collaboration and monetisation.
What does MDMP do?
MDMP is an open protocol that facilitates the creation of data sharing schemes:
- Discovery and monetisation of data and services
- Implementation of powerful, flexible governance models that can be ‘stacked’
- Management of liability and the transfer of risk based on a full audit trail
- Assurance of data quality as a result of provenance, attestation or usage
- Interoperability between heterogeneous data sharing schemes
- Accounting and settlement for ‘work done’ across and between schemes
MDMP provides a semantic model that enables the implementation of:
- A complete registry of all objects in a given data space or data sharing scheme
- A point level audit log of activity that records all prior actions
- Complex permissioning models which manage access to data and resources
MDMP gives primacy to metadata, separating it from underlying data and treating it as valuable data that needs to be captured, stored, accessed and governed
How does MDMP achieve this?
MDMP takes the generality and flexibility of a sentence-based metadata model (like RDF) and combines it with the practical requirements for a data sharing and metadata ecosystem which can happily exist and operate between multiple disparate applications. The intended purpose of MDMP is to:
- Provide a common lingua franca for describing general data spaces which allows many different resources and applications to build and manage data sharing processes more efficiently
- Provide a “mesh”, which pre-existing applications, schemes and protocols can easily interface with, to increase overall interoperability and decrease unnecessary bundling of resource provision
- Encourage the development of various toolkits to address data sharing challenges which can be more broadly adopted because they sit on a common framework
- Allow for complex “set based” workflow and permissioning. Currently, most workflow models are focused around access management only and typically assume that the author, legal owner, and controller of a data object are the same – this breaks down in many real-world data sharing problems
- Allow analysis of complex patterns within sets of (suitably permissioned) metadata. We argue that a environment that is “trustable” is often far more useful than its binary equivalents (i.e. fully “trusted” or “trustless” environments)
MDMP is designed to be a simple, technology agnostic, highly extensible open-source framework that envisages an ecosystem of one or more Metadata Service Providers who hold and manage metadata and provide services to owners and users of data.
What makes MDMP different?
The MDMP design principles set out what a universal protocol for metadata management would need to look like, if it is to be widely adopted to support greater agency and data sharing.
- Open: published, with medium term goal of making MDMP a public protocol governed by the MDMP Foundation (or a suitable established foundation)
- Distributed: assumes the existence of multiple inter-operating instances of MDMP which does not require the centralisation of data or metadata; avoids single point of failure; reduces central dependency; increases redundancy; encourages an ecosystem of competing participants to drive value by owning some, none or all of the technology stack
- Technology agnostic: operator can choose their preferred environment (including blockchain)
- Thin: allows operators to choose their own security solutions, service providers, etc.
- Universal: is not bound to – or restricted by – a primary use case or application
- Extensible: verb set (i.e. actions) can be extended to accommodate future needs
- Neutral: does not assume any primacy of login, or favour any party
- Trustable: operators define who they trust and what the rules of engagement are (e.g. using a fully “trusted” or “trustless” environment, or a risk-based framework)
- Customer focused: provides basis for end users to take control of their own data
MDMP was designed to establish an infrastructure to support the data spaces that are needed, while retaining trust.