NERVE is live across local, private and approved-cloud deployments. Book an evaluation
A common envelope with typed content.
Canonical objects share one handling envelope and carry a typed content body, so unrelated sources become comparable without losing the detail that makes each one specific.
Information model
5 min
One envelope
Every canonical object shares the same envelope. The envelope carries stable identity, acquisition and transformation provenance, handling markings, and the time fields that matter for retrieval. Because the envelope is shared, a message, a document and a sensor observation can be filtered, sorted and governed in the same way.
Typed content
Inside the envelope, content is typed. Messages carry participants and a body. Documents carry structure and extracted text. Media records carry technical characteristics and any derived text. Location observations carry a place or coordinate with a stated observation method. Generic records carry tabular rows with typed columns.
Extension namespaces
Source-specific detail is not discarded to fit the model. Fields that only make sense for one family are kept in an extension namespace attached to the object. The shared envelope stays stable, and specialist detail remains available to anyone who needs it.
Three layers
Original, Canonical and Derived are kept distinct at all times. The Original is the acquired bytes. The Canonical layer is the normalised representation. The Derived layer holds anything produced by a transformation, such as extracted text, thumbnails or optical character recognition output.
What the model does not do
Normalisation makes objects comparable. It does not resolve identity. Where the same identifier appears in two sources, NERVE records that observation and leaves the judgement about whether the same person is involved to the analyst and to CORTEX.
Key points
One envelope carries identity, provenance, handling and time for every object.
Typed bodies cover messages, documents, media, location observations and generic records.
Extension namespaces hold source-specific fields without altering the shared envelope.
Three layers stay separate: Original, Canonical and Derived.
Where this stops
Normalisation makes objects comparable. It does not merge identities: a matching identifier in two sources is an observation about the data, not proof that the same person is involved.
Related
Next
See the behaviour on your own material.
An evaluation runs NERVE against an approved sample in your environment, with the originals retained and every transformation recorded.