Tools, Implementations and Products
Knowledge Extraction Technology (KET) is implemented as a
set of tools (we call it "KETools"). They are programs written in different languages and
making together a consistent tool kit for creation of numerous data
analysis architectures and to support the knowledge extraction paradigm (see
principles in "Technology" and
"White Paper" sections).
The basis of the tool kit consists of 800+ modules. Each
module is designed to support a particular step in KET processing.
Because of their functional autonomy, modules can be used not only in
applications built on KET ideology, but also sometimes as add-ons to enhance many
other applications. For this reason, KET is licensing its modules to
software development companies interested in efficient incorporation of
AI features into their products. See details on modules and their
licensing in "Modules" section below.
Functions in KET modules can be called in a conventional
procedural way. However, a more efficient way is to use KET API that
supports calls in two scripting languages: Knowledge Extraction Language
(KEL) and SELFF. They are explained in
"API" section (see below). Here,
we will only mention that API (especially KEL) allows efficient use of
KET capabilities with minimal knowledge of KET architecture and without
any knowledge of its internal calling conventions.
Some modules of KET have server capabilities (*see
details in "Servers" section). It is especially convenient in for
implementation of architectures based on distributed computation.
A typical analysis project ends up with a discovery of a dependence
pattern, which can be presented in a descriptive way or implemented as a
program. In the latter case, the client can easily execute the pattern
for prediction, early warning, simulation, monitoring, etc. KET tool kit
contains programs that facilitate transformation of patterns into
specialized applications. It must be clear that the resulting
applications are distinct from KET modules themselves, they are
derivatives from a specific data analysis. See details in the section
Other Parts of KET Tool kit
To finish a brief description of KET's "physical
substance", we will point out that several additional parts of KET not mentioned above
also play important role in its architecture.
They are: language interpreters supporting KEL and SELFF,
special generic scripts in these languages called "knowledge elements"
and meta-utilities (schemas, templates and programs) that facilitate the
building architectural schemas under the KET paradigm.
Go to sections below to review
different types of tools and products