PET Lab - National Chiao Tung University

Cerebra Web’s semantic structure is based on a currently ongoing development of a domain model for BCI activities: a BCI ontology, which describes special concepts and properties that encapsulate the formal vocabulary and metadata definitions for processing the information generated by multimodal BCI data capture activities.
Its main design objectives are:
  • Define a minimalistic and simple abstract model for BCI activities.
  • Proposed a generic consensual knowledge about BCI activities.
  • Capture the relevant BCI metadata sets for software agents’ interoperability in a pervasive M2M environment.
  • Reuse of core and structural ontologies from different domain knowledge for specific concepts, properties and relationships.
The BCI ontology captures, in an organized way, the core structure and semantics of the following specialized BCI metadata vocabularies:
  • Extensible Data Format (XDF): is a general-purpose container format for multi-channel time series data with extensive associated meta-information stored as XML, called XDF metadata schemes. XDF is tailored towards bio-signal data (Multimodal data capture) but can easily hold da-ta with high sampling rate (like audio) or high numbers of channels (like fMRI or raw video), as well.
  • EEG Study Schema (ESS): is a XML-based specification that holds a metadata hierarchy for describing and documenting electrophysiological studies and their raw recorded data, in a format that is both machine and human readable.
  • Hierarchical Event Descriptor Tags for Analysis of Event-Related EEG Studies (HED): defines a hierarchy of standard and extended de-scriptors for EEG experimental events that provides a uniform human- and machine-readable interface, that facilitates the use of an underlying event-description ontology during EEG data acquisition, analysis, and sharing. HED tags may be used to mark and annotate all known events in an experimental session. As a classification system, HED is a folksonomy (also known as collaborative tagging), due that can be used collaborative-ly to create and manage tags for annotating and categorizing EEG-related events content. ESS is the companion specification of HED.
The vocabularies are logically grouped in two modules:
  1. BCI Metadata: based on XDF and ESS, this module groups the special and relevant concepts, relations, data properties and constraints of the information related to the BCI activities.
  2. BCI Annotations & Tags: based on HED, defines a set of hierarchical events descriptors schemes for annotations (data tagging) in specific situations for the multimodal data capture (specifically, EEG events).
Its core structure is based on the extension of the SSN ontology, W3C’s proposed framework ontology for sensor network applications. Its composition and alignment are shown in the following figure.
The BCI ontology core concepts are aligned to the Stimulus-Sensor-Observation ontology design pattern, as illustrated below.
  • Stimuli: these are the detectable changes in the environment that trigger the sensors to perform observations. BCI ontology extends the Stimulus concept by defining the HED descriptors of all EEG stimulating events as its sub-class hierarchy (which includes the following first level notions: visual, auditory, tactile, pain, taste, TMS and smell).
  • Sensors: these are the physical objects that perform observations. The design pattern makes a clear distinction between the object of sensors and the procedure of sensing. Sensors are the composite abstraction of sensing devices while the sensing procedures are the descriptions that specify how sensors should be realized and deployed in order to measure certain observable properties. Sensors can be combined to sensor sys-tems and networks. In BCI ontology, the concept of Sensor is extended by adding a BCI Device as a specialized concept of Sensing Device.
  • Observations: these are multi-dimensional objects that capture infor-mation about the stimuli, the sensors, their outputs and the spatial-temporal specification of the sensing activity. Observations are regarded as social, not physical, objects. In BCI ontology, the concept of Observation is extended to include all Sessions of BCI activities, which are related to many other concepts (a compound metadata set). All of these con-cepts are included under the BCI Metadata Module.
The BCI Semantic Data Model has been implemented using Semantic Web technologies; specifically, we use RDF/OWL as an ontology representational language.
Powered by W3C
This proposed BCI ontology, it’s still in its earliest form (a proto-ontology). We invite BCI domain experts and professionals to collaborate with us in this ongoing effort to develop a more complete, accurate and cognitive adequate semantic model for BCI activities. The further and robust development of a BCI ontology, will not only enable a consensual knowledge about BCI activities to be easily integrated into the Linked Data world, but it also allows the reusability and extensibility of its vocabulary in all kind of BCI future applications.
Last update: $2014/08/07 01:39$