KNOWLEDGE ENGINEERING USE FOR THE INTELLECTUAL SUPPORT OF MODELS’ TRANSLATION

Abstract

In the work the problem of reuse of earlier developed software models in complex systems
and their components, arising before researchers in case of necessity of transition to new modeling
tools, is considered. As a solution to this problem, a Multitranslator software environment was
developed, which made it possible to implement multilanguage translation of models' source codes
into the required format of the target modeling environment using the created translation modules.
Then, based on the Multitranslator, a client-server application was developed – a Distributed
models library, which, along with the models translation function, performed the function of their
network storage and access, providing a distributed implementation of the approach. The development
of the approach and the Distributed models library was carried out in the direction of
translation automation and resolving exceptional cases that occur during model translation
caused by insufficient input data or uncertainty in model conversion decisions that occur when
there are too many outcomes during parsing. To solve this problem, it was proposed to use an
expert system with a knowledge base. Knowledge engineering is considered as the main process of
synthesis of necessary knowledge for the knowledge base. The following sources of knowledge
acquisition during the development of the expert system are proposed: the translation module of
Multitranslator; technical documentation of input/output languages for describing models for
translation; extended and additional publications on describing these languages; experts on languages
for describing models for translation. The main stages of knowledge engineering are considered
next: defining a knowledge acquisition strategy; identifying knowledge elements; creating
a knowledge classification system; developing a detailed functional layout; pre-planning of control
transfer processes; and defining system requirements. The results obtained will allow expanding
the functionality of the Distributed models library when translating models using an expert
system and efficient processing of uncertainties that arise during translation.