Database Reference
In-Depth Information
Chapter 13
Last Words: Conclusion
Abstract We first discuss the requirements of a modern data mining system and
show that the approach presented in this topic fulfills most of them. However,
the full realization of this approach is often thwarted by principal problems in
the development of the required mathematical instruments. Especially, most of the
computational methods developed by mathematicians over the last centuries are
designed for engineering problems. We stress the differences to the requirements
for data analysis problems and encourage the development of appropriate frame-
works. Especially, control theory should play an important role here.
We will conclude by briefly summarizing the approaches to developing modern
recommendation engines described in this topic and breaking them down into seven
general requirements of a modern data mining system:
1.
Autonomous operation
: System learns automatically, no manual operation
required.
2.
Realtime operation
: System learns and decides in real time.
3.
Integration into applications
: System is embedded directly in applications.
4.
Control problem approach
: System learns through interaction, “cybernetic”
thinking.
5.
Operator description
: Mathematical formulation via operator equations.
6.
Hierarchical approach
: System uses hierarchical methods and architecture.
7.
Distributed operating principle
: System operates on a decentralized, distributed
basis.
The requirements are interdependent to some extent of course: realtime
operation, for example, requires the ability to work autonomously. And they are
not necessarily all indisputable. But they illustrate key requirements and funda-
mental trends, the use of which will ultimately lead to a new quality of data mining.
Most current data mining systems meet almost none of these requirements.
Rather than operating autonomously, they have to be operated manually, by
