Information Technology Reference
In-Depth Information
The rest of this paper is organised as follows: Section
2
describes what type of
information might be received and why it can be useful. Section
3
gives examples
of related work. Section
4
gives step-by-step examples of how the process might
work. Section
5
tries to de
ne the processes formally, as would be done for a
database. Section
6
gives some suggestions, relating the structure more closely to
nature or the physical world. Section
7
describes how this
fits in with an earlier
cognitive model and linking mechanisms research, while Sect.
8
gives some con-
clusions on the work.
2 Adding Structure to Semi-structured Data
Computers require some level of standardisation or structure, to allow them to
process information correctly. The problem is therefore how to add this structure, to
give the computer system a standardised global view over the data. Even the idea of
structure is not certain and can be different for different scenarios. Therefore,
obtaining the correct structure probably also means the addition of knowledge to the
system. As described in the related work in Sect.
3
, this type of modelling started
with relational databases (Codd
1970
), but then extended to semi-structured and
even completely unstructured information. Wikipedia
1
explains that distinct defi-
nitions of these are not clear for the following reasons:
1. Structure, while not formally de
ned, can still be implied.
2. Data with some form of structure may still be characterised as unstructured if its
structure is not helpful for the processing task at hand.
3. Unstructured information might have some structure (semi-structured) or even
be highly structured but in ways that are unanticipated or unannounced.
The introduction of random events and time elements means that the data
sources can also change (Zhang and Ji
2009
), requiring statistical or semi-intelligent
processes to recognise patterns that cannot be determined beforehand. This could
result in a different type of modelling problem than for a traditional database. For
one scenario, the designer creates a model of what he/she wishes to
nd out about
and then dynamically adds speci
rm
the model. For another scenario, the actual model itself is not known but is derived
from an underlying theory. With the second situation, not only are the model values
updated dynamically, but the model itself can change in a dynamic and unknown
way.
c data instances, as they occur, to try to con
1
Search WWH ::
Custom Search