The real-world application of Natural Language Processing (NLP) challenges for low-resource languages is growing rapidly. There has been a considerable deal of study using various ways to increase the quality of answers to these problems. This paper is devoted to automated inference from Uzbek texts using the TF-IDF approach.