WHOI ont-dev usage

One Example Usage of the MMI Device Ontology

The OceanSITES project is building SensorML (SML) files to document the datasets in the project; data is stored in NetCDF, which does not provide enough structure to include all the needed information. The SML files will provide complete metadata to describe a data file and all the processes, including devices, that produced it.   Many of the deployments in this project are complex, and a single data file may contain data from many different types of devices; some of these devices are also complex, composed of several  “sub-devices”.

The OceanSITES SML files have become very long and unwieldy, and contain information about devices that is repeated in many files. For example, every one of our deployments includes at least two SBE-37 temperature-conductivity devices; these will need to be described at least once but up to 25 times in each of about 60 SML files.

I propose to simplify these files by using a URI within the SML file to point to a description of each device; the contents of the record at that URI could be designed in several ways, but I propose to use the simplest implementation I can think of.

The external device record will hold the fields common to all instances of a particular device.  These may include most (but not all, TBD) of the fields in the ”facet list” in sections:

• Device Construction
• Functionality: Parameters, measurements, atomic phenomena
• Operational restrictions, capabilities, processes
• Operational Restrictions
• Measurement Capabilities
• Physical Processing

The SML file will contain a URI for that descriptive record, plus the details of the instances of the device that was used.  These will be details that are specific to this deployment of this individual unit, such as:

• SN
• Sample rate
• instantaneous status (calibrations),
• deployment details (e.g. x,y,z,t info).
• any deviations from the standard described in the prototype record

The details about what fields are truly universal for a device model and which are instance-level fields are still to be determined. I‘ll review our current SML files to find all the common fields for various instrument models; the facet list is not configured to make the distinction between generic  fields and instance fields.

I'm assuming that the SML content can “trump” the generic device record, which means that some of the instance fields could be included in the generic model record.