TECHNOLOGIES OF INFERENCE IN  SOFTWARE SYSTEMS

N. Ausheva; S. Shapovalova

doi:10.33842/2313-125X-2023-23-11-20

N. Ausheva National Technical University of Ukraine Igor Sikorsky Kyiv Polytechnic Institute (Kyiv, Ukraine) https://orcid.org/0000-0003-0816-2971
S. Shapovalova National Technical University of Ukraine Igor Sikorsky Kyiv Polytechnic Institute (Kyiv, Ukraine) https://orcid.org/0000-0002-3431-5639

DOI: https://doi.org/10.33842/2313-125X-2023-23-11-20

Abstract

The target of research is the implementation technology of inference. The purpose of the article is to determine the main areas of research on automatic inference, software obtained on their basis, and relevant areas of application.

Inference is a common task that is implemented in application software. The efficiency of the software systems, in particular, is determined by the performance of the built-in inference engine. Means of implementing inference shall provide optimal execution time and interaction with other components, as well as meet the requirements of the application task. The article highlights the main areas of improving the software implementation of inference engine and relevant areas of research: (i) expanding the concept and combining several paradigms of logic programming (probabilistic logic programming, defeasible inference, coinductive programming); (ii) reduction of data exchange time between software components and processing time of large knowledge bases (improving pattern matching algorithm for implementing rule-based systems, creating or expanding implementations of inference tools for integration into distributed software systems, developing new implementations of inference engines for specialized programming languages and general-purpose languages); (iii) a combination of conceptually different approaches to inference (a combination of both approaches to inference on logical and productive concepts, the integration of inference paradigms and neural network approach). The combination of software implementations of different concepts, first of all, inference engines and neural network models, gives new opportunities to artificial intelligence,

For each area of research, there are presented software tools, as well as examples of areas of their application in accordance with this concept.

Key words: inference engine, probabilistic logic programming, defeasible reasoning, rule-based systems, PRISM, ProbLog, XSB Prolog, SWI Prolog, DeepProbLog.

Downloads

Download data is not yet available.