A "template" captures a case to be found in the data set,
that is, a template is the matching mechanism for TRAKS.
Many sets of instance data in a dataset may match a template.
In order to define a set of classes connected by meaningful, we "mark" or "tag" a resource by
declaring that it is of type traks:Template such as:
<rdf:type rdf:resource="&traks;Template" />
where traks is the namespace "http://lsdis.cs.uga.edu/proj/traks#"
( the ontology -hack- is defined at:
Furthermore, for efficient matching, a subset of the template is defined as "core template".
This is done by defining that a resource is of type traks:CoreTemplate.
A resource that is "Core Template" is critical for the template to match.
Our algorithm matches first the core template nodes to filter out matches that
are not going to be relevant in the end result.
Examples are the best way to get started.
Consider this example ontology:
semdis example ontology
( touchgraph view: semdis example ontology )
Consider a scenario that captures the finding of a Terrorist that has citizenship in
a Country where a Terrorist Organization is located and such terrorist
is situated at a City located in Iraq and where a Terrorist Event has
This scenario can be defined by "marking" instances of that scenario, that is, instances
of the types formatted with italics in the previous sentence.
Such template is available here in owl format:
template_example.owl. Note how the relevant entities
to be matched in test data are marked to be of type "traks:Template", and few of them are
marked to be of type "traks:CoreTemplate" since they are critical for the scenario to make sense
and thereby they also help to make our algorithm more efficient.
1. Create a New Template through the TRAKS User Interface:
Create New Template
2. Upload a New Template by providing the URL to retrieve it,
and giving it a name to use in TRAKS web environment