I have always been trying to identify the fundamental elements of knowledge. 

I believed that once these elements are discovered, we will understand how we learn, and how we reason. We will be able to build systems that understand our environment deeply and help us utilize more knowledge in our decision-making.

I have well over 20 years of development experience mostly for Wall Street firms like Citi Corp, Bank of America Merrill Lynch, HSBC and Lehman Brothers. I developed a number of systems ranging from user interface frameworks to data mining tools.


I did my M.Sc. in Knowledge Representation Schemes a long time ago, designing a model which may be the father of the Datum Universe. I believe that the various technology elements are coming together now to enable effective knowledge processing systems for business. 


You can send me email at azmi@datumtron.com

Knowledge Representation is the core of intelligence


The Datum Universe represents knowledge using one abstract entity and one abstract relation.  These represent the minimum needed to build any kind of structure. The DatumTron API provides an interface to this structure.  The API tutorial shows how to import, query and mine a full relational database.  The current implementation is in .NET but I am planning to support more platforms.

My vision is that with the Datum Universe and DatumTron, we can: 

  • Build systems that are able to understand and learn knowledge at the deepest level needed.  
  • Understand how the brain represents knowledge, since the neural network of the brain looks very much like a datum graph.
  • Study knowledge itself using the Datum Universe Graph.  We can identify common subgraph and macro structures associated with higher more complex types of knowledge.  We can then, quantitatively evaluate and analyze knowledge.


I am looking for partners who share this vision and who would like to advance these ideas into all areas of database and artificial intelligence applications.  My wish list looks like this:

  • Port the DatumTron API to more platforms,
  • Build Datumtronic servers that utilize classical hardware to compete with existing database servers,
  • Explore new architectures like IBM's Neurosynaptic chips to build Datumtronic AI servers. 
  • Build Knowledge "Understanding" agents using Natural Language Parsers like the Stanford University parser to be the new Google.  
  • Build financial trading systems that can detect emerging market patterns in real time.

Experience

Founder of DatumTron

Datumtron

Knowledge Representation Technology

Ashraf Azmi  (Ash)

Partnerships