POSITION STATEMENT FROM OASIS CLASSIFICATION OF EVERYDAY LIVING (COEL) TC

For presentation at Data Privacy Controls and Vocabularies, a W3C Workshop on Privacy and Linked Data April 17-18 2018, Vienna, Austria.

The COEL-TC (https://www.oasis-open.org/committees/tc_home.php?wg_abbrev=coel) manages the development of an OASIS Open Standard that implements a privacy-by-design framework for the collection and processing of behavioural data. The standard provides a means to improve interoperability for organisations that seek to create working privacy-by-design systems. Applications built using COEL deliver data privacy and data self determination for individual citizens and consumers, in a way that is fully auditable and easy to check legal compliance.

There are four key technical insights that have been used to build COEL as the basis for the transparent use of dynamic data for personalised digital services, IoT applications where devices are collecting information about identifiable individuals and the coding of behavioural data in identity solutions:

[1] For privacy-by-design applications, we have found that a minimum requirement is to pseudonymise personal data at source and then maintain a strict and auditable separation of different data types with aligned roles & responsibilities for all actors involved.

[2] We have defined new data types based on the expertise of COEL-TC members in digital capture of human behaviourial data. The well defined data types in COEL standard are one of the core means to deliver interoperability for users of the standard. All behavioural data are defined as event-based packets. Every packet is connected directly to an individual and can contain a summary of the consent they provided for the processing of the data. This provides a means to store and process context specific consent using e.g. Kantara CISWG-TC protocols (https://kantarainitiative.org/groups/ciswg/) or other consents standards.

[3] The Classification of Everyday Living is a holistic, hierarchical taxonomy that sits at the heart of the COEL Standard. For applications that require granular data on everyday human behaviour, we believe the COEL taxonomy provides a unique and extensible knowledge base that can deliver semantic harmonisation for personalised services. The first version of this asset already delivers a high level of harmonised knowledge, and the COEL-TC framework provides a structured means for this to develop in years to come to a broader and richer ontology derived from use cases. An interactive visualisation of the current COEL JSON artefact has been created by Coelition (https://coelition.org/) to showcase this asset base (https://coelition.org/business/resources/visualising-life/). We believe that the combination of the COEL taxonomy of human behaviours, and the event-based data type and data handing protocols, provide a universal template for data portability. We have consciously kept the knowledge that is encoded in the COEL taxonomy independent of the technical infrastructure used for the COEL standard.

[4] Privacy-by-design applications cannot be achieved with a pure technical solution. In parallel with the well-defined data types and taxonomy, the COEL approach also defines roles & responsibilities for actors who want to use the COEL standard in privacy-by-design implementations. The COEL Standard describes simple interface specifications that help enforce the separation of roles and provide system-level interoperability.