Home Contact

about us
technology
white paper
FAQ
services
projects
tools
download
 

Welcome to knowledge extraction home!

“Knowledge Extraction Tools, LLC” (KET, LLC) is a software development and data analysis company providing services based on a proprietary “Knowledge Extraction Technology".  KET, LLC creates and customizes decision support software that typically relates to a niche business: medical diagnostics, financial evaluation, security early warning systems, etc.

“Knowledge Extraction Technology is a “assumption free" approach to automatic construction of models from empirical evidence (which may include data, semantic nets, conceptual frame systems, knowledge bases and results of knowledge acquisition from experts).

The methodology is “assumption free" in the sense that models (and even types of models) describing data are not assumed in the beginning of the analysis, they are being built or identified in the process of analysis.

Approaches used in KET reflect paradigm shifts occurred in information technology during the last three decades:

bullet

Shift from storage-oriented understanding of data management to the communication oriented approach. Regardless of the formal representation of the analyzed input, we find interpretation of sources a virtual database (or knowledge base) very fruitful

bullet

Shift in programming and artificial intelligence from structural to object-oriented to agent-based paradigm. Unification of different types of sources and their treatment as open systems helps to solve problems, which are seemingly serious, but with a shift of interpretation are ephemeral

If to be brief and non-technical, we can characterize our approach as based on three principles:

bullet

Unification and "meaning". All types of sources including relational data, free text expert descriptions and numerous other types are treated under one representational umbrella. Instead of interpretation of sources as the focus of analysis, they are understood as messages about "object of investigation" behind the scene. This introduces and provides the formal definition of "meaning".

bullet

Consistency. Sources may contain errors and irrelevant parts. Statements of experts may be biased or incorrect. Generalized concept of inference helps to extract a consistent part of what is available.

bullet

Open system ideology. The analysis process may go beyond what is given. The tools may ask questions back. In contrast to simple systems, which can provide answers or say "don't know", our tools generate questions to the point: they focus the experts (or a search system)  on promising directions, which may provide useful extension of input

Principles themselves live in IT for a long time, but to base a technology on them consistently is not a trivial task.

Basic principles of the “Knowledge Extraction Technology” are described in the “White Paper on KE Technology” section (see also "Frequently Asked Questions"). Physically it is implemented as a big tool kit "KETools" containing 800+ modules.

For a more complete list of available products and services, see “Services" and "Tools" sections. For examples of projects completed in the past, see "Completed projects" section.

For those who prefer a non-technical overview of ideas and motivation behind the technology, we included a little story describing what KET is about (see "KET Story" in our KET Archive),

 

  New feature !!

 A new KET module was introduced recently to support interaction with "Mathematica 8" system.

 

  News !!

KET, LLC joined BioMed Content Group, Inc. in initiative of using AI agents to facilitate work of physicians and educators.

Copyright 2002-2006, Knowledge Extraction Tools, LLC. All rights reserved