Management and Future Technologies | Volume 2, Issue 3 | № 007

• Column 4 – questions correctly detected; • Column 5 – questions missed (classified as non-questions); • Column 6 – declaratives misclassified as questions; Tabel 1. The results of the CRF model are provided as well. Interrogative It's not a question F1 Score Number of sentences sentence Interrogativ e sentence It’s not a Interrogati question ve sentence It’s not a question Literature 11659 958 11 32 10658 97,8 Anatomy 10565 0 0 24 10541 0,0 Biology 4757 211 11 25 4510 92,1 Botany 2717 217 12 27 2461 91,7 Religion 6079 195 14 24 5846 91,1 Physics 12078 1474 13 22 10569 98,8 Geography 6237 236 14 19 5968 93,5 Law 1964 102 18 20 1824 84,4 Social Sciences 421 9 2 4 406 86,1 Informatics 2512 70 12 12 2418 85,2 Economics 520 0 0 5 515 0,0 Chemistry 3566 313 21 20 3212 93,9 Cinema 668 1 0 1 666 100 Mathematics 7921 569 11 23 7318 97,1 Textbook Reader 3382 408 12 34 2928 94,7 Language Textbook 11020 275 22 26 10697 92,0 Agriculture 1209 0 0 12 1197 0,0 Short Story 10529 697 14 26 9792 97,2 Laws and Resolutions 4360 0 0 24 4336 0,0 Text types 260

Fine Arts 3673 25 4 12 3632 85,1 Novel 1896 120 21 19 1736 85,7 Art 9570 189 12 23 9346 91,5 Politics 1222 6 1 2 1213 79,6 Sport 2565 71 5 26 2463 82,2 Traditional Medicine 421 5 1 2 413 87,9 History 4619 318 15 24 4262 94,2 Technology 796 66 12 8 710 87,1 Medical Science 5490 201 14 23 5252 91,6 War 1084 34 6 11 1033 80,2 Zoology 4348 416 16 24 3892 95,4 total 137848 7186 294 554 129814 89 Conclusion To date, virtually no models have been built for detecting question marks in Uzbek texts, making the model proposed here especially relevant. We developed a CRF-based approach to the NLP problem of identifying interrogative sentences in Uzbek. Applying the model to a corpus of 1 ,642 ,860 words and 137 ,848 sentences across 30 text categories yielded an F1-score of 89 %. As future work, we plan to tackle Uzbek question detection with more advanced AI models in pursuit of even higher accuracy. References [1] M. A. , M. N. , A. D. Mahmudov N., O’ZBEK TILI ME"YORLARI (Punktuatsiya), vol. 1. Tashkent: Zamin nashr, 2021. [2] J. Pokrywka, “Punctuation Prediction for Polish Texts using Transformers,” Feb. 2024. doi: 10.48550/arXiv.2410.04621. [3] P. Żelasko, P. Szymański, J. Mizgajski, A. Szymczak, Y. Carmiel, and N. Dehak, “Punctuation Prediction Model for Conversational Speech,” Feb. 2018, pp. 2633–2637. doi: 10.21437/Interspeech.2018-1096. [4] M. S. Sharipov, H. S. Adinaev, and E. R. Kuriyozov, “Rule-Based Punctuation Algorithm for the Uzbek Language,” in International Conference of Young Specialists on Micro/Nanotechnologies and Electron Devices, EDM, 2024, pp. 2410 – 2414. doi: 10.1109/EDM61683.2024.10615061. [5] M. Bayraktar, B. Say, and V. Akman, “An analysis of English punctuation: the special case of comma,” International Journal of Corpus Linguistics, vol. 3, Jul. 1998, doi: 10.1075/ijcl.3.1.03bay. [6] D. Hardt, “Comma checking in Danish,” 2001. 261

[7] U. Salaev, E. Kuriyozov, and C. Gomez-Rodríguez, “SimRelUz: Similarity and Relatedness scores as a Semantic Evaluation Dataset for Uzbek Language,” in 1st Annual Meeting of the ELRA/ISCA Special Interest Group on Under-Resourced Languages, SIGUL 2022 - held in conjunction with the International Conference on Language Resources and Evaluation, LREC 2022 Proceedings, 2022, pp. 199 – 206. [Online]. Available: https://www.scopus.com/inward/record.uri?eid=2-s2.085138700420&partnerID=40&md5=bf476cd74317f06577dd0548c5c600d6 [8] A. M. Abdurashetona and I. O. Ismailovich, “Methods of Tagging Part of Speech of Uzbek Language,” in Proceedings - 6th International Conference on Computer Science and Engineering, UBMK 2021, 2021, pp. 82 – 85. doi: 10.1109/UBMK52708.2021.9558900. [9] A. M. Abdurashetona and U. Mokhiyakon, “Software Features and Linguistic Features of Uzbek Synonymizer,” in Proceedings - 7th International Conference on Computer Science and Engineering, UBMK 2022, 2022, pp. 171 – 175. doi: 10.1109/UBMK55850.2022.9919447. [10] B. Mengliyev, S. Shahabitdinova, S. Khamroeva, S. Gulyamova, and A. Botirova, “The morphological analysis and synthesis of word forms in the linguistic analyzer,” Journal of Language and Linguistic Studies, vol. 17, no. 1, pp. 558 – 564, 2021, [Online]. Available: https://www.scopus.com/inward/record.uri?eid=2-s2.085103797274&partnerID=40&md5=8a4052419f721c3c734f8fe1c48984ec [11] K. Madatov, S. Bekchanov, and J. Vicic, “Dataset of Karakalpak language stop words,” Data Brief, vol. 48, 2023, doi: 10.1016/j.dib.2023.109111. [12] M. Sharipov, J. Mattiev, J. Sobirov, and R. Baltayev, ‘Creating a Morphological and Syntactic Tagged Corpus for the Uzbek Language’, CEUR Workshop Proceedings, vol. 3315, pp. 93–98, 2022. [13] D. Mengliev, E. Akhmedov, V. Barakhnin, Z. Hakimov, and O. Alloyorov, “Utilizing Lexicographic Resources for Sentiment Classification in Uzbek Language,” Jan. 2023, pp. 1720–1724. doi: 10.1109/APEIE59731.2023.10347765. [14] K. Madatov, S. Bekchanov, and J. Vicic, “Dataset of stopwords extracted from Uzbek texts”, Data in Brief, vol. 43, 2022. 262

WORKING WITH PARALLEL CORPORA: APPROACHES AND TOOLS IN COMPUTATIONAL LINGUISTICS Amirkulov Marufjon Tashkent State University of Uzbek language and literature amirkulov.edu01@gmail.com Abstract : This article is a conceptual study that explores the concept of parallel corpora, the methods for their creation and analysis, as well as the tools used in the field of computational linguistics. It examines processes such as segmentation, tokenization, tagging, and alignment in bitext corpora created in Arabic, English, Uzbek, and other languages. The paper reviews examples from previous research, such as machine translation, lexical-semantic analysis, and machine learning. Parallel corpora are presented as vital resources for automating language projects, building terminological databases, and conducting applied linguistic research. Keywords: segmentation, text collection, tokenization, tagging, alignment, cleaning Introduction A parallel corpus is a specially prepared collection of texts in which each segment is linked, sentence by sentence or even phrase by phrase, with corresponding translations in one or more other languages, enabling direct comparison and computational processing [1]. Parallel corpora play a significant role in linguistic research and machine translation systems, as they allow the observation and modeling of units in one language through examples of their translation into another [6]. However, the lack of data and resources hinders the achievement of accurate results, and therefore, the effectiveness of a machine translation system largely depends on the quality of the corpus used [2]. The creation of parallel corpora, the extraction of knowledge from them, and their application to practical systems represent a major direction in computational linguistics. These processes involve a range of technological tools, models, and algorithms. This paper is conceptual in nature and does not present original experimental results but synthesizes and reviews approaches, tools, and methodologies from the existing literature. The process of creating parallel corpora Creating a parallel corpus from scratch is a multi-stage process. In general, the process proceeds as follows: A. Data collection: First, texts translated between the required languages and domains are collected. This is done either by importing from existing digital resources (e.g., official document databases) or by crawling the web. At this stage, the problem of identifying document pairs also arises—for instance, finding the original and translated versions of a book, or detecting multilingual versions of a web page. B. Conversion to a common format and cleaning: Files from various sources may come in different formats (PDF, DOC, HTML, txt). These need to be converted into a unified text format (e.g., UTF-8 encoded txt or XML). During this process, unnecessary elements (images, table borders, special characters) are removed, and general encoding issues are resolved (e.g., converting HTML entities into actual characters). C. Document-level alignment: 263

If the corpus is document-based, the document in one language must be paired with its translated version. This is done by analyzing file names, metadata (title, date, author), or internal links. In some cases, automatic document alignment uses the bag-of-words method: words in documents are treated as vectors, and the cosine similarity is used to find the closest pairs. Once a document match is found, their identifiers (or file names) are added to the list of parallel pairs. D. Segmentation (splitting into sentences): Segmentation is performed within each matched document. This is done separately for each language based on its punctuation rules. Segmentation tools may be used at this stage. As a result of segmentation, a sequence of sentences is obtained for each document, and they are numbered. E. Sentence alignment: Now, sentences within the matched documents are linked to each other. The simplest method is the Gale-Church algorithm, which aligns sentences based on their length (in characters or words). For more accurate results, dictionary-based alignment is used (e.g., HunAlign), where frequently occurring common words (e.g., nouns) are identified across the parallel texts, and this information is combined with sentence length. Modern methods rely on embedding-based alignment (e.g., vecalign, BERTalign): each sentence is mapped to a feature space, and matching pairs are found based on vector distance. At the end of this stage, an alignment score may be assigned to each sentence pair, allowing low-quality alignments to be filtered out later. F. Word alignment (optional): If the goal is to extract lexical information from the corpus, word alignment can be performed for each sentence pair. This is usually done with lightweight tools such as the IBM Model 2 or FastAlign. As a result, a list of aligned word pairs is generated for each sentence pair. This data can be stored in a separate file or integrated into the corpus. G. Tagging: Additional linguistic annotations can be added to the parallel corpus. For example, the lemma (dictionary form) of each word, its syntactic role, semantic labels (e.g., NER – Named Entity Recognition), and so on. Pre-trained models such as Universal Dependencies treebanks, spaCy, or Stanza can be used for this. Adding annotation enables more complex queries and analyses later (e.g., searching only for translations of verbs, or studying how certain syntactic constructions are translated). H. Storage as database or file format: The final parallel corpus is prepared for distribution to users. At this stage, the corpus may be shared as a set of simple files (e.g., TMX or txt files in a zip archive) or imported into a multifunctional platform (e.g., SketchEngine, OPUS). Often, large corpora are loaded into a database (e.g., NoSQL databases, formats like Corpus Workbench) and made searchable via a web interface. These stages may differ slightly depending on the type of corpus being created. 264

Fig. 1. Parallel Corpus Creation Proces For example, in a document-based corpus, step 3 is critical, while in a sentence-based corpus, step 3 may already be resolved at the source (e.g., translation memory). In domain-specific corpora, the cleaning stage can be more complex. For instance, in a medical corpus, formulas and chemical expressions require special attention. Nevertheless, the overall approach follows this sequence. Alignment algorithms and models The main scientific challenge in working with parallel corpora is alignment—the task of finding equivalent segments and units. Over the past 30 years, computational linguistics researchers have proposed many alignment algorithms: 1) Gale and Church (1993): This was one of the first effective methods proposed for aligning sentences in parallel texts by length. The translated sentence length is assumed to be proportional to the original sentence length. The number of characters in each sentence is counted, and a normal distribution model is built to predict one language's sentence length from the other. Then, dynamic programming is used to find the most probable alignment path. The Gale and Church algorithm is fast and language-independent (it does not require a dictionary), which is why it was widely used in early corpus projects [3]. 2) Church and Gale (1991) – Token-based method: This approach performs alignment based on the number of full stops ("."), assuming that the number of periods in English and French texts align. However, this method is limited and not applicable to many languages. 3) CAligner, Vanilla: These are modified versions of Gale–Church, used in projects like JRCAcquis. They are based on similar principles but differ in implementation details (e.g., how segment weights are calculated). 4) HunAlign (Varga, 2005): This popular alignment tool uses a two-stage approach. First, it performs length-based alignment without a dictionary. Then, from the initial alignments, it automatically generates a bilingual dictionary (collecting matched word pairs). In the second stage, this dictionary is used to refine the alignment and resolve difficult cases. 5) BleuAlign (2010): This method uses a machine translation model for alignment. First, a raw translation model is trained for a language pair. Then, for each chunk, it searches for the one whose translation produces the highest BLEU score [4]. This reframes the alignment task as: “Which chunk, when translated, yields the best translation?” As a result, semantically aligned segments are found. BleuAlign is effective on small datasets but less efficient for large corpora. 265

6) BERT-based aligners (2019–...): In recent years, with the emergence of multilingual neural networks (e.g., BERT, XLM-R), a new alignment approach has gained popularity. A multilingual model (e.g., mBERT) generates contextual embeddings for each sentence, and cosine similarities between vectors are used to identify matching sentences [4]. BERT-based alignment is relatively slow, but with modern GPUs, it is now applied to large corpora. 7) Doc2Vec alignment: Some studies propose using Doc2Vec embeddings for entire documents to perform similarity-based document alignment [5]. This is useful for the task of identifying document pairs (as discussed in step 1). For example, if a novel and its translation exist, their Doc2Vec vectors will be very close, while other books will be distant. Thus, even matching book pairs from a large library becomes possible. 8) Corti analytical alignment: Some systems use linguistic rules and logical analysis (e.g., parse trees) to align texts. For example, comparing parallel dependency trees, and considering trees with similar structures to be aligned. These methods are complex and less commonly used but have academic value. It should be noted that today, word alignment has become an internal component of machine translation models. In neural network-based translation, explicit word alignments are difficult to extract, but attention mechanisms provide approximate alignment matrices. Some studies have even achieved unsupervised alignment, generating cross-lingual embeddings from monolingual corpora, then aligning them using methods like MUSE, to identify word correspondences. This is already being applied in dictionary construction. NLP tools and architecture Working with parallel corpora involves software systems with multi-layered architectures. Fig. 2. Parallel Corpora with Multi-layered Architecture 266

Let’s examine large corpus-based platforms such as Sketch Engine or Corpus Manager: 1) Database: Corpus segments and alignments are stored in a specialized database that supports a powerful search system (e.g., Corpus Query Language – CQL). These databases are not typical SQL databases but are built on inverted index structures, since they need to support fast word-based searching. 2) Server layer: This component receives the user’s query (e.g., “kitob 𝑢𝑧 – book 𝑒𝑛”), searches the corpus database, and prepares the results. If sentence-level alignment is used, the result will return each found Uzbek sentence and its English translation on the same line. 3) Frontend (user interface): In a browser or GUI application, the user views the results; the matched segments can be highlighted in different colors (e.g., the searched word in red). In model training scenarios with parallel corpora, the architecture is different. For instance, a neural machine translation (neurMT) model reads the parallel corpus in formats such as CSV or TFRecord, splits it into mini-batches, and feeds it to a GPU. When the corpus is large, shuffled shards are prepared, the data is pre-cleaned (e.g., sentences longer than 50 words are removed), and so on. Tagging processes (POS tagging, parsing) are often organized as a batch processing pipeline. A distinct aspect is the use of translation memories (TM) and CAT tools (Computer-Assisted Translation tools). In these systems, the parallel corpus (TM) architecture is optimized on the server side for fast segment-level search. On SDL Trados Server, segments are hashed, and as soon as a translator enters a segment, the system retrieves the most similar one from the database using its hash. This is another example of parallel corpus search architecture—except here, the user sees similar translations in real time as they enter text. Working with parallel corpora brings together multiple subfields of computational linguistics: linguistic knowledge (e.g., dictionaries, grammar analyzers) improves alignment; statistical and probabilistic models (IBM models) uncover matches; and modern neural networks utilize contextual associations. At the foundation of all this lies the parallel corpus—a large-scale database that serves as a fundamental resource for both scientific research and practical applications. Conclusion Monolingual and parallel corpora form the foundation of modern linguistics and language technologies. In particular, parallel corpora serve as a bridge between different languages. They illuminate the regularities of the translation process, fuel machine translation and other multilingual systems, and provide invaluable material for language learning and comparative analysis. This paragraph reviewed the types of parallel corpora (document-based and sentence-based, domainspecific), their areas of application, structure (formats, alignment, segmentation, metadata), and their characteristics as illustrated by the most well-known corpora. The referenced scientific sources and examples serve to present the topic of parallel corpora in accordance with the demands of research in theoretical, applied, and computational linguistics. It should be especially noted that working with parallel corpora is a continuously evolving field: thanks to neural networks, deeper alignment and translation capabilities are emerging, and more and more languages are becoming digitally connected. Therefore, parallel corpora will continue to maintain their strategic importance at the intersection of linguistic theory and applied computational linguistics. 267

References [1] I. Simeon, “Parallel corpora and multilingual dictionaries,” Filologija, no. 38–39, pp. 0–0, 2002. [Online]. Available: https://hrcak.srce.hr/173311 [2] D. Banik, A. Ekbal, and S. C. Satapathy, “Fuzzy influenced process to generate comparable to parallel corpora,” ACM Trans. Asian Low-Resour. Lang. Inf. Process., just accepted, June 2023. https://doi.org/10.1145/3599235 [3] https://quod.lib.umich.edu/cgi/t/tei/tei-dx?type=pointer&value=SACSAL [4] F. Forgac, D. Munkova, M. Munk, and L. Kelebercova, “Evaluating automatic sentence alignment approaches on English-Slovak sentences,” Scientific Reports, vol. 13, no. 1, p. 20123, 2023. [5] S. Paun, “Parallel text alignment and monolingual parallel corpus creation from philosophical texts for text simplification,” in Proc. 2021 Conf. of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop, June 2021, pp. 40–46. [6] F. Zanettin, “Parallel corpora in translation studies: Issues in corpus design and analysis,” in Intercultural Faultlines, Routledge, 2017, pp. 105–118. [7] A. Eisele and Y. Chen, “MultiUN: A multilingual corpus from United Nation documents,” in LREC, May 2010. 268

ADVANCING EDUCATIONAL METHODS THROUGH ICT INTEGRATION Shohida Yusupova; Ogabek Matyoqubov Department of Information Technology Urgench Branch of Tashkent University of Information Technologies named after Muhammad alKhwarizmi gratifikus@gmail.com Abstract: The article discusses the problems and prospects of organizing distance learning in higher educational institutions of Uzbekistan in the context of the COVID-19 pandemic. Particular attention is paid to the integration of information and communication technologies and mobile technologies into the educational process, which has significantly transformed teaching methods. The study is based on the experience of introducing electronic test materials into the educational process of the Urgench branch of the Tashkent University of Information Technologies. The use of interactive educational tests made it possible to activate independent cognitive activity of students, as well as to increase the efficiency of knowledge control and organization of independent work. The methods of developing tests using the LaTeX package and its acrotex library are described, which ensures the creation of various types of tests that can take into account the individual characteristics of each student. The authors emphasize the importance of using electronic test materials for monitoring and training, and also analyze the impact of modern technologies on the development of students' professional skills. The presented results demonstrate successful experience in integrating distance technologies and emphasize their potential for further use in education. Keywords: education systems, distance education, mobile technologies, cloud technologies, knowledge control, test, LaTEX, moodle.Introduction At the beginning of 2020, Uzbekistan, along with the rest of the world, faced an unexpected external factor that brought about a qualitatively new stage in the development of the state, society, economy and education. World medicine turned out to be unprepared to fight the new virus, and therefore the only more or less effective means of preventing the rapid development of the epidemic was the maximum restriction of contacts between people, their self-isolation. In this situation, the adequate response of the country's education system was the transfer of educational organizations to a remote form of organizing the educational process. As a result, for the first time in its history, the country's higher education system found itself in a fundamentally new situation when all educational activities, including education management at all levels, were transferred to a remote format, involving the organization of training using distance learning technologies. Unlike all previous modernizations and reforms, this situation is characterized, firstly, by the coverage of the entire education system of the country at all levels of its functioning; secondly, by the unexpectedness, the absence of any special preliminary preparation for such a transition; thirdly, the simultaneous transfer to a qualitatively different format of training. In this context, it should be noted that in developed countries, the organization of distance learning using information and computer communications has a long history. Accumulated experience indicates two main directions of these processes: 269

1) online education, which involves distance learning of formal higher education via the Internet; open education, which means the creation of open educational resources for informal education and possible use in formal education. The Resolution of the President of the Republic of Uzbekistan dated September 26, 2017 No. PP-3289 "On measures to further improve the system of training, retraining and advanced training of public teachers" provides for the development of an educational and methodological base for advanced training courses, improving the organization of training courses in both online and offline formats. Particular attention is paid to the introduction of innovative scientific achievements, modern pedagogical and information and communication technologies, as well as distance learning and the expansion of independent learning in the education system. In connection with the above-mentioned resolution, numerous distance courses have been created in Uzbekistan. However, the procedures of external quality assurance common in the world, such as professional and public accreditation, examination of educational programs, institutional examination and public accreditation, are still rarely used in Uzbekistan. Nevertheless, there is a tendency to increase the number of audits and certifications of management systems of organizations in the field of education. Therefore, the strategic choice of the majority of Uzbek participants in the educational process is based on systematic internal work on quality management aimed at continuous improvement of education [1]. The rapid development of mobile and cloud technologies, especially popular among the generation of "digital natives", has actualized a large number of scientific studies that convincingly prove that the integration of mobile technologies into the educational process can significantly transform it. For example, it is noted that the introduction of mobile technologies contributes to the targeted consideration of the individual characteristics of each student and, accordingly, the organization of autonomous learning. In addition, they provide students with freedom, that is, the ability to work outside the classroom at any convenient time, make it possible to quickly diagnose problems in the learning process, create a mobile interactive learning environment and provide instant feedback, lead to the emergence of new forms of presentation of educational materials, help create new formats of interactive tasks based on the use of augmented reality and geolocation applications [2]. In the context of reduced hours for classroom work, teachers in higher education institutions must organize their teaching activities in such a way as to motivate students to study such disciplines as mathematics and develop professional skills. The problem of finding optimal didactic possibilities of modern information technologies is becoming relevant. 2) Analysis Of Sources Many researchers, such as N.G. Maloshonok, J.S. Sahlin, A. Tsertsidis and M.S. Islam, who study issues related to online education, note a positive relationship between students' involvement in achieving educational goals and interacting with teachers and classmates and teachers' use of the Internet and multimedia technologies. Online education is a new type of activity at this stage of development, and therefore it is of interest to both teachers and students, thereby stimulating changes and expansion of the practice of its use. The attitude towards this educational project is quite contradictory. On the one hand, there is an optimistic point of view, considering this learning mechanism as productive and meeting the needs of the modern era. A number of studies are associated with forecasts about the prospects for the development of MOOC pedagogy, based on an 270

analytical study of existing learning outcomes [3]. On the other hand, experts also express pessimistic sentiments, expressing distrust of this method of teaching in general [4] and pointing to the lack of development of general pedagogical principles for the mass implementation of this technology [5]. Among the disadvantages of e-learning, the most important are the problem of selforganization of students and the rigidity of the material taught. Disadvantages also include ignoring the individual characteristics of the student, limitations in obtaining practical skills, dependence on technical means, low quality of the material taught and control over the level of knowledge acquired [6]. The emergence of additional cognitive load. Cognitive load is understood as the amount of mental memory required by the user to achieve the goal. According to I.G. Zakharova, working with interconnected web pages involves performing a parallel series of actions: first, the student has to assimilate the material read; second, keep in mind the logical chain of judgments when moving from one hyperlink to another. As a result, the context of the problem being solved is expanded [7]. To improve the efficiency of students' academic performance, it is necessary to shift the center of gravity of the educational process towards greater independence of students, to develop effective technologies for teaching students’ rational methods of independent activity. In their studies, scientists argue about the impact of training using modern information and communication technologies on the development of students' independent work skills and higher learning outcomes, that didactic tools in combination with ICT-based tools contribute to achieving better results. It is believed that elements of interactivity, in particular educational tests, can be presented as an effective didactic method of remote work for university students. Knowledge control is also one of the important elements of organizing students' remote work. In the context of an electronic educational environment, the issues of the effectiveness of using testing as knowledge control are becoming extremely relevant, many researchers are engaged in the issues of constructing control tests and their reliability. The issues of organizing automatic assessment of students' work are also relevant. Testing as a type of control of students' educational and cognitive activity is considered in many didactic works. In modern didactics, a test acts as a convenient and reliable tool for diagnosing and monitoring the educational process. Issues of testing technologies related to substantive and methodological filling. There are different opinions among teachers about the use of testing in studying a discipline: from a complete refusal to elevating it to the rank of a pedagogical panacea. As a rule, underestimation of testing is based only on insufficient awareness of teachers [8]. Scientists believe that the development of interactive test materials solves several important problems: • promotes the activation of independent cognitive activity of students, while ensuring their mobility, • provides systematic control over independent work of students, • expands the competencies of the teacher in the field of ICT technologies, • increases the level of developed electronic educational and methodological complexes, • at the same time frees up the teacher's time, which is important in the context of the intensification of the work of a higher education teacher [2]. 271

The aim of the study is to improve the efficiency of organizing students' distance learning in the process of studying the disciplines Calculus 1, 2, included in the training programs implemented in a technical university, mainly by the Urgench branch of the Tashkent University of Information Technology named after Muhammad al-Khwarizmi, and through the use of electronic test materials of an educational and control nature in the educational process, containing elements of interactivity. Materials and methods of research Main research methods: • analysis of scientific literature on the problems of developing testing technology in education, the problem of managing students' distance learning and the possibilities of using modern information and communication technologies in education; • analysis and generalization of regulatory documents related to the educational process at the university; • analysis of practice and experience in the field of studying mathematics in technical universities; • analysis of the choice of environment for the implementation of test materials. Discussion And Results The essence of the complete mastery system used in distance education is expressed in the following stages of work: • Formulation of diagnostic learning objectives. • Development of complete mastery standards. • Development of standards and tests to check the degree of mastery of educational material. • Differentiation and individualization of students based on existing (before the start of work on the complete mastery system) indicators. • Varying the time of training and learning. A noticeable increase in the share of time for independent work. • Development of new educational materials based on the modular principle. • Development of assignments for self-monitoring for all modules studied. • Development of tests for pedagogical control of readiness for each module and for the entire course. • Organization of independent work of students, during which the teacher collaborates with students to resolve educational difficulties that arise from time to time in individual students. Correction of knowledge based on the results of self-monitoring. • Testing [2]. To create tests, the authors chose the acrotex library, which is part of the LaTeX package. Let's list the reasons why we chose acrotex. With the acrotex package, it is possible to organize quite a few types of tests, the creation of which in MOODLE is quite labor-intensive, sometimes impossible. The authors described the technology for creating such materials in their works [9]. It should be noted that an important advantage of the acrotex package is the ability to check the correctness of the solution and display the number of points scored. After that, the student can save the file with the solutions and send it via LMS MOODLE to the teacher to familiarize themselves with the results of his work. Since the file has the .pdf format, there is no way to "fake" the answer. • the LaTeX package is free; 272

• LaTeX is the de facto standard for entering mathematical expressions not only in printed works, but also in web publications, such as in LMS Moodle; • LaTeX allows you to easily enter complex mathematical expressions (fractions, matrices, tensors, etc.); • when using acrotex, entering an answer (even a numeric one) provides more options for the user. For example, you can enter not only numbers with a decimal point, such as 0.333, but also ordinary fractions, such as 11/13, and even special expressions, such as pi/2, sqrt(2), etc. in the input field. • you can use mathematical functions in the answer, for example, sin(pi/4), ln(3/2), 7^2+12^4, etc. the answer can be symbolic, such as 5*x^3-15*x+12, ln(sin(2x)+cos(3x)) etc. the input of tasks is more convenient than constructing a test window in visual environments. It is much faster to type simple text, which is a LaTeX file, than to manually place individual graphic primitives on the screen, and LaTeX itself will neatly place everything on the page [9]. Educational test materials as didactic tools that promote better assimilation and memorization of educational material. The compiled tests have the property of variability due to the inclusion of tasks of different levels of complexity, which ensures differentiation of the learning process and allows you to build an individual learning trajectory for each student. The general structure of the test materials contains: 3) Theoretical material presented in the form of an interactive electronic lecture. 4) Examples of key problems with a fully substantiated solution. 5) Questions for self-assessment. 6) Educational interactive tests 7) Educational and control tests for independent completion by the student (http://moodle.ubtuit.uz/mod/quiz/attempt.php?attempt=115105&cmid=40173). Let us describe the technology of creating test materials using the example of the topic "Integral calculus of a function of one variable". The topic covers the following issues: antiderivative; indefinite integral; method of replacing a variable and the method of integration by parts; definite integral; applications of definite integrals to geometry and mechanics. The experimental platform for the study was http://moodle.ubtuit.uz/ "Urgench branch of the Tashkent University of Information Technologies named after Muhammad al-Khwarizmi", on the basis of which these materials were tested by the authors in the practice of teaching full-time and part-time students. The review of the topic begins with the study of theoretical material - lectures. Reports contain definitions, basic theoretical provisions and evidence of the main facts. Then the student is offered detailed examples of typical problems and test questions for self-assessment. After studying the theory of the topic and test questions, the student must solve the test. The tests compiled by the authors and used in their work are conventionally divided into training and control. Training tests are aimed at practicing the main concepts, basic skills of action, algorithms for the subject under consideration. The tasks included in the training test have different levels of difficulty. The test consists of theoretical tasks, tasks aimed at memorizing the main features of formulas or concepts, basic tasks on the topic, tasks requiring the restoration of missing movements or establishing the correct sequence of actions. 273

Solving the training test is aimed at identifying gaps in the assimilation of the educational material. Having solved the problem and entered the answer in the required form, the student immediately sees the result of his actions in points. If an incorrect answer was entered, the test provides the ability to view the errors and get acquainted with the detailed correct solution. Another goal of the training test is to familiarize the student with the process and the assessment system. The training test is presented in three versions of different levels of difficulty. He solution of a training test is to identify gaps in the assimilation of the educational material. Having solved the problem and entered the answer in the required form, the student immediately sees the result of his actions in points. If the wrong answer is entered, the test allows you to view the errors and find the correct solution. Another goal of training tests is to familiarize the student with the process and the assessment system. The training test is presented in three versions at different levels. The next step in studying the topic is solving a control test. Such a test also performs an educational function, so it can be called a training and control test. Testing does not include checking errors and finding the correct solution. When completing control tasks, the student immediately sees the test results in points, but does not see where exactly he made a mistake. The authors offer two options for using control tests in the learning process. The first is control tests as a tool and method for organizing independent homework; the second is as a test in a classroom [10]. Control tests are compiled in many versions for each topic, each of which is presented in three levels of difficulty. The student has the opportunity to choose the level of difficulty of the test, having previously independently assessed the level of his readiness. We will indicate the types of tasks that we used when creating tests and give examples of these tasks. 1. Closed form of test tasks. Such test buildings do not imply entering your own answer. The test taker must choose the correct answer or several answers from the proposed ones. This type of tasks also includes tasks to establish a correspondence or the correct sequence of actions. The authors used the following tasks when constructing the test: • - with a choice of the only correct (or incorrect) answer from those offered, • - with an alternative choice, • - with a choice of several correct (or incorrect) answers, • - to establish a match, • - with a choice of the most correct answer, • - tasks requiring the restoration of the correct sequence of actions (the answer in such problems is the correct set of numbers or letters indicating the next answer or action) or tasks in the process of solving which it is necessary to restore the missing elements. 2. Open form of test tasks. This type of task implies that the test taker enters his answer in a symbolic version. The following were used: • - tasks requiring a numerical answer; • - tasks requiring an answer in the form of a symbolic expression; • - tasks requiring the addition of an answer in free form with restrictions on the answer (numerical, symbolic). The test materials and the tests themselves are designed as a file in .pdf format. 274

Here is an example: Calculate the indefinite integral: ∫ sin(√𝑥 + 𝑎) 𝑒 √𝑥 √𝑥 𝑑𝑥 Writing this task in LaTEX: \(\int\dfrac{\sin\left(\sqrt{x}+a\right)\mathrm{e}^\sqrt{x}}{\sqrt{x}} \) Question "Drag into text" This test is a set of empty cells for missing text or sentences in the form or words of the test. The test participant determines the correct set and sequence of words in the sentence, selects the necessary words from the proposed options and drags them with the mouse into the empty cells. In this case, you can add unnecessary options that complicate the task. The authors use these types of questions in this article [11]. Brief instructions 1. 1) Go to the main page of the course you need (click on the course name in the list of available courses). 2) On the course page, in the upper right corner, click the gear icon "Action Menu" and on the page that opens, select the tab "Question Bank" -> "Questions". 3) From the drop-down list with the names of the Question Bank categories, select the category to which the question you are creating will belong. Click the "Create a new question..." button. 4) In the window that opens, select the question type "Drag and drop into text". Click the "Add" button. 5) In the "Question Title" field, type the initial part of the task wording to make it easier to find in the general Question Bank. In the "Question Text" field, write the question text like: "[[1]] [[2]] [[3]] [[4]] [[5]]" or "Text [[1]] text [[2]] text [[3]] etc.", then in the "Choice Options" section, enter the correct words for the labels marked with numbers 1, 2, 3, 4, 5, etc. Add extra options if necessary. Set a score for correct completion of the task or leave the default score. Be sure to enable answer shuffling (check the box). 6) If necessary, add comments in the "General Feedback on the Question" field or set penalty percentages and hints for each new attempt to complete the task. 7) After clicking the "Save Changes and Continue Editing" button, you will be able to preview the created question. 8) Save the created question – the “Save” button at the bottom of the page. 9) Use the magnifying glass icon to view the added task and make adjustments if necessary. In order to use the embedded JavaScript, you need to run Adobe Acrobat locally. To do this, you must first save the .pdf file to disk, and then run Adobe Acrobat. Trying to run Adobe Acrobat on the server is useless, since third-party macros are prohibited in browsers for security reasons. The acrotex package is connected with the command: \usepackage[xetex,unicode=true]{exerquiz} The next step is to connect the dljslib package to use JavaScript. \usepackage[setSupport,ImplMulti,limitArith,indefIntegral]{dljslib} To activate the task, you need to click the “start” button; to see the result, click the “stop” button. After this, the “Answer” field will display the result in the form “Correct: X of Y” (see Fig. 1 and 2). 275

Fig. 1. An example of a training task with symbolic input and the ability to view the solution before entering the answer. Fig. 2. An example of a training task with symbolic input and the ability to view the solution after entering the answer and pressing the “Stop” button. Conclusion The authors attempted to create materials that form a holistic system of knowledge, skills and abilities on a given topic. The materials were tested on the site http://moodle.ubtuit.uz/ "Urgench branch of the Tashkent University of Information Technology named after Muhammad alKhwarizmi". A lot of preparatory work was done to select theoretical and problem material. Test materials are designed to partially compensate for the lack of classroom communication between the teacher and the student, and therefore the authors place special emphasis on the problem material. The tasks in the tests are selected in such a way that without serious study of the material, the student cannot complete the task. The created electronic resource contains elements of interactivity. The electronic resource differs from existing online services primarily in that its purpose is not only control, but above all training, which involves deep and conscious study of the material in the smallest details. Its peculiarity is that with the help of the existing tests, various learning skills and algorithms are not only checked, but also practiced. The mathematics course at a technical university is quite extensive. Providing students with a variety of educational multivariate materials on each topic at least within one semester requires a lot of time and labor. At present, the authors have developments only on individual topics. Therefore, the authors believe that at present, the main methods of research and diagnostics of the effectiveness of the materials being introduced are questionnaires, surveys, and conversations with both students and mathematics teachers. As shown by a survey conducted among students and teachers, about 86% claim that the use of such test materials helps the student to study the subject and achieve the required educational level. 276

More than 92% of respondents noted that the use of the proposed materials stimulates regular student study, since the completion of educational test assignments was organized in deadline mode. Among the advantages of this resource, teachers noted the possibility of an individual approach to learning, the ability to track the progress of the educational process, as well as quickly and objectively control this process. Also, as a plus, it was noted that the resource does not depend on the availability and speed of the Internet connection. Based on the positive experience of testing the developments presented in the article, the authors intend to continue working on the creation of electronic learning resources, including thematic interactive test materials. References [1] S. B. Yusupova and S. Z. U. Davletboyev, “Organization of Monitoring in the Example of LMS Moodle”, IEEE XVI International Scientific and Technical Conference Actual Problems of Electronic Instrument Engineering (APEIE), Novosibirsk, Russian Federation, 2023, pp. 1550-1556, doi: 10.1109/APEIE59731.2023.10347622. [2] V. S. Avanesov. “Application of tasks in test form in new educational technologies”, School technologies. – 2007. – No. 3. – pp. 146-163. [3] M. Ebben & J. S. Murphy. “Unpacking MOOC scholarly discourse: a review of nascent MOOC scholarship”, Learning, Media and Technology, 2014, 39(3), 328–345. https://doi.org/10.1080/17439884.2013.878352 [4] J. Baggaley. “MOOC rampant”, Distance Education, 2013. Vol. 34. No. 3. P. 368–378 [Electronic resource]. URL: http://dx.doi.org/10.1080/01587919.2013.835768. [5] M. Clarà & E. Barberà. “Learning online: massive open online courses (MOOCs), connectivism, and cultural psychology”, Distance Education, 2013, 34(1), 129–136. https://doi.org/10.1080/01587919.2013.770428 [6] M.V. Klimenskikh, N.A. Korepina, A.S. Sheka, O.S. Windecker. “Peculiarities of perception of distance learning by university students and teachers”, Modern problems of science and education. - 2018. - No. 1. - P. 41. URL: http://science-education.ru/ru/article/view?id=27421 [7] E. A. Akolzina. “The use of electronic educational resources in the learning process: advantages, disadvantages”, Gaudeamus, 2013, No. 2 (22). URL: https://cyberleninka.ru/article/n/ispolzovanie-elektronnyh-obrazovatelnyh-resursov-vprotsesse-obucheniya-dostoinstva-nedostatki (access date: 12/10/2023). [8] Sh. B. Yusupova, F. A. Masharipova. “Application of the type of test tasks “tasks with the choice of one most correct answer”, ILM SARCHASHMALARI. 3.2021. 181-185 b. [9] T. A. Bondarenko, G. A. Kameneva, A. L. Anisimov. “The use of information and communication technologies as a condition for organizing independent work of a student”, Problems of modern pedagogical education, 2019, No. 62-1. URL: https://cyberleninka.ru/article/n/ispolzovanie-informatsionno-kommunikativnyh-tehnologiykak-uslovie-organizatsii-samostoyatelnoy-raboty-studenta (date of access: 05/28/2023). [10] A. L. Anisimov, T. A. Bondarenko, G. A. Kameneva. “Development of modern test materials for organizing students’ independent work in studying higher mathematics using the LaTeX package”, Perspectives of Science and Education. – 2019. – No. 2 (38). [11] https://edu.susu.ru/edususudocs/ru/help-for-teachers/testing/test-questionstypes/create-question-drag-and-drop-into-text/sozdanie-voprosa-na-peretaskivanie-v-text.pdf. 277

JURNALISTIK FAOLIYATNI MONITORING QILISH VA JURNALISTLAR REYTINGINI BAHOLASH JARAYONLARINING MODELLARI VA AXBOROT TIZIMINI DOLZABRLIGI Omonullayeva Farangiza Vatan University universiteti “Raqamli texnologiyalar” kafedrasi o‘qituvchisi Annotatsiya: jurnalistika jamiyat hayotining eng muhim ijtimoiy institutlaridan biri sifatida tezkor axborot uzatish, jamoatchilik fikrini shakllantirish va demokratik qadriyatlarni mustahkamlash jarayonida beqiyos ahamiyat kasb etmoqda. Axborot oqimlarining globallashuvi, internet va raqamli platformalarning keng tarqalishi natijasida jurnalistik faoliyatning mazmuni, shakli va uslublari tubdan o‘zgarib bormoqda. Hozirgi kunda jurnalist faqat axborot yetkazuvchi emas, balki auditoriya bilan interaktiv aloqada bo‘ladigan, tahlilchi, ijtimoiy jarayonlarni baholovchi va ularni yoritish orqali ularning rivojlanishiga ta’sir ko‘rsatuvchi sub’ekt sifatida qaralmoqda. Kalit so‘zlar: jurnalistika, interaktiv, axborot, monitoring, axborot tizimi. Kirish XXI asrda jurnalistika jamiyat hayotining eng muhim ijtimoiy institutlaridan biri sifatida tezkor axborot uzatish, jamoatchilik fikrini shakllantirish va demokratik qadriyatlarni mustahkamlash jarayonida beqiyos ahamiyat kasb etmoqda. Axborot oqimlarining globallashuvi, internet va raqamli platformalarning keng tarqalishi natijasida jurnalistik faoliyatning mazmuni, shakli va uslublari tubdan o‘zgarib bormoqda. Hozirgi kunda jurnalist faqat axborot yetkazuvchi emas, balki auditoriya bilan interaktiv aloqada bo‘ladigan, tahlilchi, ijtimoiy jarayonlarni baholovchi va ularni yoritish orqali ularning rivojlanishiga ta’sir ko‘rsatuvchi sub’ekt sifatida qaralmoqda. Mazkur jarayonlarda jurnalistlarning professional faoliyatini samarali tashkil etish, ularning ish sifati, ishonchliligi, ta’sirchanligi va ijtimoiy mas’uliyatini ob’ektiv mezonlar asosida baholash masalasi dolzarb ilmiy muammolardan biri hisoblanadi. An’anaviy yondashuvlarda jurnalist faoliyatining samaradorligi odatda sub’ektiv baholarga asoslangan holda ko‘riladi, natijada reytinglar va monitoring natijalari ishonchlilik jihatidan yetarli darajada kafolatlanmaydi. Shu bois jurnalistik faoliyatni monitoring qilish, jurnalistlar reytingini aniqlash va bu jarayonni avtomatlashtirishga qaratilgan axborot tizimini ishlab chiqish ilmiy-amaliy nuqtayi nazardan katta ahamiyatga ega. So‘nggi yillarda xalqaro tashkilotlar tomonidan axborot erkinligi va media bozorining rivojlanishini baholashga qaratilgan turli reyting va indekslar ishlab chiqildi. Masalan, Reporters Without Borders tashkiloti tomonidan e’lon qilinadigan Press Freedom Index, IREX tomonidan tayyorlanadigan Media Sustainability Index, shuningdek, Freedom House tashkilotining hisobotlari mamlakatlardagi jurnalistika erkinligi va sifatiga baho berishda keng qo‘llaniladi. Ammo mazkur indekslar asosan davlat miqyosida erkinlik va sharoitlarni o‘lchaydi, jurnalistlarning individual faoliyatini, ularning jamiyatdagi rolini aniqlashga qaratilmagan. Shu bois, individual jurnalist reytingini shakllantirish, ularning faoliyatini monitoring qilish va xalqaro standartlarga mos ravishda baholash ilmiy va amaliy tadqiqotlar uchun yangi yondashuvdir. Bunda zamonaviy sun’iy intellekt, data mining, sentiment-analiz va media monitoring algoritmlarini qo‘llash dolzarb ahamiyat kasb etadi. Bu nafaqat jurnalist faoliyatini 278

ob’ektiv baholashga yordam beradi, balki global axborot makonida jurnalistlarning professional o‘rnini aniqlash imkonini ham beradi. O‘zbekistonda so‘nggi yillarda amalga oshirilayotgan keng qamrovli islohotlar jarayonida axborot siyosati va media erkinligi yangi bosqichga ko‘tarildi. Prezidentimiz tomonidan ilgari surilgan “Raqamli O‘zbekiston – 2030” strategiyasi, shuningdek, davlat boshqaruvining ochiqligini ta’minlashga qaratilgan tashabbuslar jurnalistika faoliyatining ahamiyatini yanada oshirdi. OAV faoliyati ustidan jamoatchilik nazorati kuchaygani sayin, jurnalistlarning mas’uliyati ham ortib bormoqda. Shu nuqtayi nazardan, jurnalistlarning professional faoliyatini monitoring qilish, reytingini aniqlash va ular o‘rtasida sog‘lom raqobatni shakllantirish zarurati mavjud. Bu, bir tomondan, jurnalistlarning faoliyatini ob’ektiv baholash imkonini bersa, ikkinchi tomondan, jamiyatda sifatli, haqqoniy va tezkor axborot maydonini yaratishga xizmat qiladi. Bugungi kunda O‘zbekistonda jurnalistlarning reytingini aniqlashga oid yagona tizim mavjud emas. Reytinglar ko‘proq ijtimoiy tarmoqlardagi obunachilar soni, maqolalar soni yoki auditoriya qamrovi kabi soddalashtirilgan mezonlar asosida shakllanadi. Bu esa jurnalistning kasbiy malakasi, axborotning haqqoniyligi, tahliliy salohiyati va ijtimoiy ta’sirini to‘liq aks ettirmaydi. Demak, ilmiy asoslangan baholash modelini ishlab chiqish zaruriyati mavjud. Jurnalistik faoliyatni monitoring qilish bo ‘yicha ayrim tadqiqotlar mavjud bo‘lsa-da, ular ko‘pincha kontent tahliliga yoki auditoriya xatti-harakatlarini o‘rganishga qaratilgan. Jurnalist shaxsiy faoliyatining samaradorligini aniqlash uchun esa kompleks yondashuv, ya’ni ko‘p mezonli modellashtirish talab etiladi. Masalan, jurnalistning auditoriyaga ta’sirchanligi, ijtimoiy tarmoqlarda faolligi, professional etikaga rioya qilishi, axborot manbalarining haqqoniyligi kabi ko‘rsatkichlarni umumlashtirish asosida reyting tizimini ishlab chiqish lozim. Shuningdek, bunday tizimni axborot texnologiyalari yordamida avtomatlashtirish jurnalist faoliyatini tezkor, xolis va samarali monitoring qilish imkonini beradi. Natijada, jurnalistlarning reytingi nafaqat OAV tahririyatlari, balki davlat organlari va jamoatchilik uchun ham ochiq bo‘lgan ko‘rsatkichga aylanadi. Jurnalistika insoniyat tarixida axborotni yetkazishning eng asosiy vositalaridan biri bo‘lib kelgan. Ammo raqamli texnologiyalar va internetning jadal rivojlanishi natijasida jurnalistik faoliyatning ko‘lami kengaydi, uning mazmun-mohiyati tubdan o‘zgardi. Hozirgi kunda jurnalist faqatgina matn yoki maqola yozuvchi emas, balki multimediya kontent yaratuvchi, ijtimoiy tarmoqlarda faoliyat yurituvchi, axborotni tezkor uzatuvchi va jamoatchilik fikrini shakllantiruvchi sub’ekt sifatida namoyon bo‘lmoqda. Bunday sharoitda jurnalistlarning faoliyatini ob’ektiv baholash va monitoring qilish murakkab masalaga aylanmoqda. Chunki an’anaviy baholash mezonlari (maqolalar soni, bosma yoki elektron OAVda e’lon qilingan materiallar hajmi) jurnalistning professional malakasi, auditoriyaga ta’sirchanligi, axborotning haqqoniyligi kabi muhim jihatlarni to‘liq qamrab olmaydi. Bugungi kunda mavjud bo‘lgan reyting tizimlari ko‘pincha ijtimoiy tarmoqlardagi obunachilar soni, “layk” va “ko‘rishlar” statistikasi kabi soddalashtirilgan ko‘rsatkichlarga asoslanadi. Ammo bunday yondashuv jurnalistning professional darajasini, ijtimoiy mas’uliyatini va jurnalistik mehnatning jamiyat taraqqiyotiga qo‘shayotgan real hissasini aks ettirmaydi. Natijada jurnalistlarning obro‘-e’tiborini xolisona baholashda sub’ektivlik yuzaga kelmoqda. 279