CX2 Commons

CX2 Commons is an idea: building owners voluntarily making anonymised data available for research and education under controlled terms.

Buildings produce data that researchers and students rarely get access to, because it sits in closed systems. With a shared data layer and a shared model it becomes possible to share data in a way where the building owner decides what, with whom and for how long. Data is protected. Knowledge is shared.

The problem is not a lack of data. A modern building measures plenty. The problem is that each building's data is described in its own way, so a data set from one building cannot simply be used alongside one from another. Once the buildings are described with the same model, most of that work disappears, see Semantic modelling with Brick Schema.

Who decides what

The building owner chooses which buildings and which points are contributed, and for how long, and can withdraw the contribution again. Access is not granted by CX2 alone. The idea is a small access council with a research partner, a representative of the building owners and an independent professional, who assess applications against fixed criteria and keep a record. Access is time-limited, tied to a specific project and a responsible researcher, and the data is deleted when the project ends.

Anonymisation in three layers

Structural: building, owner, address, vendor and point names are replaced by Brick classes and sequence numbers. The model still says what a point is, without saying where it is.

Temporal: time stamps are shifted or aggregated, so patterns of use cannot be tied to a calendar.

Statistical: for small rooms and single-person offices the data is aggregated or left out, because occupancy can be inferred from CO2 and temperature.

The honest caveat is that anonymisation costs utility. Research into fault detection needs raw time series, while energy benchmarking can live with aggregates. How much is lost at each level has not been measured, and it is one of the first things that needs to be settled.

What it builds on

Commons requires no new collection. It builds on what CX2 Open already does in the building, and adds three things: a contribution module that anonymises before anything leaves the building, a catalogue of buildings as anonymous models without identity, and an access-controlled layer with logging and automatic deletion. Data is handed over in open formats, so it can be read without CX2.

Requires external funding and has not been started. Neither the access council, the anonymisation levels nor the partners have been decided, and the description here is a proposal.