Information Technology Reference
In-Depth Information
Fig. 1.3.
The implementation of Ramon Llull's
Ars Magna
by S. Abbott and Y. Dambergs.
Source: [185]
Thus computational linguistics begun with the introduction and implementation
of Leibniz's universal dictionary. It is noteworthy that this method is still of inter-
est today. One example comes from 1961, when for the purposes of a machine
translation system, Margaret Masterman, a former student of Ludwig Wittgen-
steins, defined a network of 15,000 words based on compounding 1,000 simple,
basic concepts [70]. In 1975, Roger Schank reduced the number of primary ac-
tions that had to be executed when using computational linguistics to just 11
[125]. According to his concepts, the transformation of high-level concepts into
simple, basic ones, had to use two different phrases synonymous to one another.
When considering methods of computational linguistics, it is worth remember-
ing that the system generally used in it must allow the transformation of high-level
(complex) concepts into concepts (words) of a lower level. Such a system must
also have a constructive component enabling the construction of complex concepts
from correctly combined simple concepts, but this enhancement of the system,
treated as a certain change of the basic rules, must be optional, not obligatory.
All the methods mentioned above were based on logical foundations. They used a
logical apparatus for their operation, starting with Porphyry's tree down to formal
ontologies, exemplified by Aristotelian top-down approaches. However, let us re-
member that Aristotle himself preferred the use of a bottom-up method for empirical
data analysis. In 1858, William Whewell extended the definition of the top-down
method, making it more precise and dedicating it to natural science [149]. In 1865,
John Stuart Mill proposed conjectures and defined the final conditions for creating a
