Information Technology Reference
In-Depth Information
re-structure the trees that they get added to. Each tree should try to represent some
sort of distinct entity and it would be desirable to have larger trees, as this indicates
a greater level of coherence. The structure must also be as stable as possible
however and so we could try to always add to a base set of concepts, so that the
base always has the largest count values. Therefore a triangular structure is realised,
with respect to count values, where the base has the largest count, narrowing to the
branches. If this basic principle is broken, it might be an indication that the structure
is incorrect. Additions to an existing tree should include additions from the base
upwards when possible, with new concepts creating new branches if required. It
should
'
extend
'
the existing tree along the whole of one of its branches.
2.2 Structure Justi
cation
An earlier paper (Greer
2011
) gave a slightly philosophical argument that if two
concepts are always used together, then at some level they are a single entity. This
is a very general rule not related to any particular application, but describes how
any sort of entity can be important based on its relevance to the scenario. Consider
then the following made-up scenario: There is a farm with a fence in a
field.
A sheep comes up to the fence and jumps over it. Sensors in the field record this and
send the information to the concept base. The concept base recognises the sheep
and the fence objects and assigns them to be key concepts in the event. With our
existing knowledge, we would always assign more importance to the sheep, but if
we had never encountered either object, maybe the sheep and the fence would be
equally important to ourselves as well. The scenario continues, where a cow comes
up to the fence and jumps over it, then a chicken comes up to the fence and jumps
over it. In this case, the fence now becomes the main and key concept. Without the
fence,
of the scenarios can occur. A count of these concepts would give the
fence the largest total, again suggesting that it is the key concept. The process to
combine these scenarios might then compare these stored events and decide that a
concept tree with the fence at its base would be the most stable. This process is
described further after the related work section, where the addition of existing
knowledge is also suggested, to add a natural ordering to things.
'
none
'
3 Related Work
The related work section is able to include topics from both the information pro-
cessing and AI areas. After introducing some standard data processing techniques
and structures, some intelligent methods, relating to nature in particular, are
described. It would also be an important topic for problems like data management
in the business or online worlds, for example Blumberg and Atre (
2003
) or Karin
et al. (
2012
). While concepts are the main focus of interest, combining service
Search WWH ::
Custom Search