Mining for Meaning

I specialize in implementing innovative new applications of data mining and machine learning techniques.

My current project, Metadata Analytics, is described in several recent conference presentations  and articles.

Enterprises are accumulating massive stores of enterprise IT information such as schemas and software code -- "metadata." The metadata describes the IT systems, but it is nearly impossible to understand it in terms of real-life business concepts. IT departments want very much to capture the business semantics, but the only people who can do it are those who understand both IT and the business, and these are the most expensive resource of all.

Computers get faster and store more data every year. At the same time, academia is producing practical new and effective new algorithms for classifying, organizing, and discovering patterns in large amounts of data. These techniques are ideal for connecting metadata to semantics.

This can be compared to the use of data mining for analyzing gene expression data: mapping genes--the software code of biological systems--to the functionality in which they are expressed--the real world semantics of biological systems.

The time is right for Metadata Analytics, the application of well-known mining techniques to the automated semantic classification of enterprise metadata.