DETECTING HOMONIMY BY MEANS OF THE NAIVE BAYES CLASSIFIER

Abstract

One of the relevant issues of a natural language processing is word sense disambiguation.
Homonyms are considered as an important element of determining the meaning of a word. Methods based on machine learning play a special role in solving this problem. Naive Bayes classifi er is one of the important machine learning methods. When eliminating homonymy between diff erent and grammatically similar groups of words in the Uzbek language, the Naive Bayes classifi er diff ers from other methods in its simplicity and speed. This classifi er is one of the most popular multi-class classifi cation algorithms, and depending on the data in question, any of the 3 types of Naive Bayes algorithms (Gaussian, Polynomial, Bernoulli) can be used. This article scrutinizes the processes of using the classifi er to identify homonymy between grammatically similar groups of words in the Uzbek language.