Information Technology Reference
In-Depth Information
formally. This might be the most
'
intelligent
'
part of the process, as a re-join can be
compared to a self-heal or
fixing process. The non-bio systems would typically not
do this and therefore continue to a more chaotic state.
5 Formal Speci
cation
The concept tree idea, for a concept base, has a restricted construction process. It is
based on a frequency count with a very strict rule about relative count sizes. It
might therefore be possible to define the construction process more formally and
even bring some standardisation or normalisation; where other similar techniques,
such as Navigational Database or NoSql, are not able to. The following sets of
declarations might be useful to standardising the process and bring added order to
the structures. Initial tests have con
rmed some of these rules, but are not variable
enough to test all of the possible scenarios. Because the rules are more of a logical
argument than a mathematical proof, they are listed as groups of points.
5.1 General
1. A concept tree can represent different types of entity. The entity however should
be a whole concept. For example, it might be a single object in the real world, or
a single action. Therefore, the base concepts in any tree are the ones that would
be used
first in any scenario.
2. Tree structures require that every child node has a count that is the same or less
than its parent. This should always be the case if the linking integrity is
maintained, unless branches are allowed to re-join.
3. Whenever possible, the process would prefer larger trees for the following
reasons:
(a) A larger tree has more meaning as a general concept and gives added
con
dence when reasoning over its group of nodes.
(b) A larger tree gives more coherence to the concept base.
(c) Larger trees mean less of a trend towards a chaotic structure.
4. Normalisation would like each concept to exist only once and so, also for this
reason, the whole process tries to
find what the main concepts are and place
them as base concepts to trees. As with traditional databases, if a concept exists
somewhere only once, then it only needs to be updated in one place. This is
dif
cult or even impossible however, for every scenario:
(a) If the concept gets used for different contexts, then its meaning and relation
to other concepts changes slightly, when it needs different link sets.
(b) For a distributed system over a large area, it might simply not be practical
to have the concept at one place only and be able to
find it.
Search WWH ::
Custom Search