Author Archive

Database Driven Virtual Reality

March 1, 2008

As a brief history, I have been involved in building database driven information systems since the mid 1980’s, have done much work on semantic networks and knowledge representation, participating in research on early systems like KL-ONE and Candide.  Merging with database technology, these systems then became semantic data models, and have now matured to the current day ontology languages such as OWL.  We were looking for a better way to represent complex information in a way that the meaning can be processed by a computer, in contrast to the traditional relational databases which are very weak on semantics.  I have always been trying to use these to create information collections for various departments and colleges here at UF, and have built numerous systems which are currently on-line.  Applications I am involved with include digital libraries, ontology-based simulation, natural language processing, eLearning, and 3D/VR.

The goal in every case is to represent knowledge in the domain within an underlying ontology database, or what I call an ontology management system, which is a database built on ontology concepts (the data are modeled using a formal ontology language).  I have built the Lyra OMS as my current platform for this work.  Knowledge objects in the OMS can automatically transform to presentation environments, so we generate our eLearning modules automatically for a student presentation by mapping the content in the OMS to a display based on desired styles of presentation.

In 3D/Virtual Reality applications, I only began working in this area about 2 years, we have successfully demonstrated how the OMS can store the objects in the VR scene. We built a prototype using the Java Xj3D environment that couples our OMS to a VR rendering engine.  There is much potential here, because in this environment the scene geometries can be directly integrated with other  knowledge, so we no only have the shape of an object (which is what you normally find in a VRML file), but also the meaning of the object, vast taxonomies of object categories, dynamic behavior (through integration with our ontology-based simulation environment), any other associated knowledge, and the ability to name objects using multilingual, natural language expressions.

So since we can project objects from the OMS to a presentation, I figured we’d have no problem exporting our VR scene from the OMS into Second Life, but we now know this is impossible using today’s SL technology.  And I’m dreading trying to implement our dynamic simulation models in LSL (the scripting language used in SL).  I am about to begin exploring alternative platforms that might make my life easier.  Anyone know of any?

http://edis.ifas.ufl.edu   (our ancient agricultural extension publication database)

http://ees-diglib.ad.ufl.edu/WCC (our more recent digital library work)

http://cbc.at.ufl.edu  (our new eLearning site for a national training course in crop protection)

http://orb2.at.ufl.edu/Aymara  (Aymara on the Internet, our language training program)

http://orb.at.ufl.edu/ObjectEditor (my general purpose site listing my projects)