Browsing by Author "Ismagulov N."

Now showing 1 - 2 of 2
  • Thumbnail Image
    ItemOpen Access
    Development of a system for generating test questions and answers for a given text
    (2022) Ismagulov N.
    Currently, there is a tendency to increase the volume of documents containing complex text structures. Professional text proofreaders provide their services to correct errors in the text for not a small amount of money. A very large number of people every day around the world, who are closely connected with science or education. write a large volume of articles and, accordingly, each and all of them Should be written without errors and in the correct version. In order to automate this process, it was necessary to develop an algorithm based on the methods of analysis and correction of the input text. For high-quality text synthesis, it iS necessary to use machine learning technologies that require deep knowledge and understanding in this area. Among the many machine learning algorithms Hunspell algorithm seems to be one of the best to solve this issue. The essence of this algorithm is to bring all the words contained in the text to the original format. Thus, this work is based on multilevel segmentation of errors of Kazakh Language text from the Internet or by manual user input. It is worth noting that due to technological progress, the main source of linguistic research is social networks, which is a critical problem due to the dubious fidelity of texts. The Main purpose of this work was the formation and development of a spell-checking algorithm for the Kazakh language based on the existing Hunspell algorithm for English. As a result, the Hunspell algorithm was studied, where the methods of extracting the base of the word, as well as the ways of aggregation of the normal form were analyzed.
  • Thumbnail Image
    ItemOpen Access
    QUESTION ANSWERING SYSTEM UPON UNIFIED LANGUAGE MODEL AND EVALUATING PERFORMANCE OF DATASETS
    (СДУ хабаршысы - 2023, 2023) Ismagulov N.
    Abstract. Present days require automation and optimization in simple but urgent tasks. It is granted to use opportunities of technologies and science in order to work efficiently and to stay productive. In this paper, I seek to understand opportunities and drawbacks of the publicly available datasets, such as SQuAD, TriviaQA, Natural Questions (NQ). QUAC, NewsQA. It is vital to choose a suitable dataset in order to create a system with better performance. Specifically, the paper proposes an automatic question creating system that uses state-of-the-art Natural Language Processing (NLP) - Unified Language Model (UniLM). The question generating algorithm was verified using best datasets, and it has shown noteworthy results - questions generated were logical and correct. This study is important for teachers, teacher assistants, to save time writing test questions and spend it for more important duties.