Term-Driven Classification of Low-Resource Mathematical Documents in Uzbek Language

This article describes an algorithm for classifying mathematical documents in the Uzbek language that relies on the use of terms with different meanings. For each section of mathematics (discrete mathematics, probability theory, mathematical analysis, linear algebra, geometry, and number theory), a list of keywords is created with weights from 1 to 3 indicating their significance. The algorithm calculates the overall score of the text based on the terms found and assigns the document to the topic with the highest value. To improve accuracy, some common words can receive a “penalty” depending on their frequency in different topics. Testing on 600 texts showed that the accuracy and recall rates are <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$94-100 \%$</tex>, indicating the high efficiency of the approach without using machine learning. In addition, the authors conducted a fairly detailed comparative analysis of existing alternative works. In addition to this, authors conducted research for looking for relevant scientific works, which shows high actuality of the current work.