Machine learning methods for identifying destructive texts in the Kazakh language

Milana Bolatbek, Moldir Sagynay, Shynar Mussiraliyeva, Zhastay Yeltay

2024

Closed access Metadata (CSL-JSON)
DOI 10.1145/3701100.3701121
Data confidence 85 high
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

In the digital age, society is faced with an increasing number of destructive messages on the Internet, including insults, racism, violent extremism and national extremism. This paper presents a machine learning-based approach to classify destructive texts in the Kazakh language, collected from social networks. During the study a text corpora has been accumulated on social networks: YouTube, Vkontakte, Telegram. The study uses advanced text vectorization methods, including Bag of Words, TF-IDF, and Word2Vec, to preprocess text data. Our findings demonstrate that Logistic Regression outperforms other models, achieving an accuracy of 86%. These results contribute to the development of automated systems for identifying destructive content, promoting safer online communication."