Renat Akhmedov
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Renat Akhmedov has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
- Vitalii Bezkorovainyi
Institute Information Technologies in Economics, Kyiv National Economic University named after Vadim Hetman,Informatics and Systemology Department,Kyiv,Ukraine
3 shared publications - Vasily Derbentsev
Kyiv National Economic University Named After Vadym Hetman
2 shared publications Kyiv National Economic University named after Vadym Hetman
1 shared publication- Mykola Bondarchuk
Kyiv National Economic University Named After Vadym Hetman
1 shared publication - Tetiana Danylchenko
Kyiv National Economic University Named After Vadym Hetman
1 shared publication Василь Дербенцев
1 shared publication
Latest Publications
- Open Access
- 17 December 2024
- Published by Centr for Sustainable Development in Смарт-економіка, підприємництво та безпека
- Vol. 2 (2) , 59-70
EVALUATING CUSTOMER EXPERIENCE IN E-COMMERCE: MULTILINGUAL SENTIMENT ANALYSIS OF USER REVIEWS USING TRANSFORMER MODELS
This study addresses the challenge of multilingual sentiment analysis in e-commerce, with a focus on Ukrainian and Russian book reviews. We propose a hybrid framework based on transformer architectures that accounts for the linguistic complexity of Slavic languages and the significant class imbalance often present in customer feedback data. Using a dataset of approximately 70,000 user reviews from the Ukrainian online bookstore Yakaboo, we evaluate four model variants based on the XLM-RoBERTa architecture and compare their performance to a monolingual Ukrainian RoBERTa baseline. The classification task is formulated as binary: identifying low-rated reviews (1–3 stars) versus high-rated ones (4–5 stars), with the minority class comprising only 4–5% of the dataset. Our best-performing model — combining partial fine-tuning, a deep classifier, and focal loss — achieves a macro F1 score of 0.73 and an F1 score of 0.47 for the minority class, outperforming the baseline (0.66 and 0.37, respectively). The application of oversampling and focal loss proved effective in mitigating class imbalance. Beyond technical performance, the findings underline the practical utility of accurate sentiment detection for e-commerce platforms. Effective identification of negative feedback enables companies to address customer dis- satisfaction, refine product offerings, and inform personalized marketing strategies. This research contributes to advancing multilingual NLP in low-resource settings and provides a scalable solution for real-world sentiment classification in morphologically rich languages. The scientific novelty of this research lies in the integration of multilingual transformer architectures with adaptive classifiers and class imbalance mitigation techniques, specifically tailored for real-world e-commerce review data in morphologically rich languages. The proposed framework demonstrates how domain-specific fine-tuning and architectural customization can significantly enhance sentiment classification performance in low-resource language settings.
- 17 December 2024
- Published by Centr for Sustainable Development in Смарт-економіка, підприємництво та безпека
- Vol. 2 (2) , 59-70
Open AccessEVALUATING CUSTOMER EXPERIENCE IN E-COMMERCE: MULTILINGUAL SENTIMENT ANALYSIS OF USER REVIEWS USING TRANSFORMER MODELS
This study addresses the challenge of multilingual sentiment analysis in e-commerce, with a focus on Ukrainian and Russian book reviews. We propose a hybrid framework based on transformer architectures that accounts for the linguistic complexity of Slavic languages and the significant class imbalance often present in customer feedback data. Using a dataset of approximately 70,000 user reviews from the Ukrainian online bookstore Yakaboo, we evaluate four model variants based on the XLM-RoBERTa architecture and compare their performance to a monolingual Ukrainian RoBERTa baseline. The classification task is formulated as binary: identifying low-rated reviews (1–3 stars) versus high-rated ones (4–5 stars), with the minority class comprising only 4–5% of the dataset. Our best-performing model — combining partial fine-tuning, a deep classifier, and focal loss — achieves a macro F1 score of 0.73 and an F1 score of 0.47 for the minority class, outperforming the baseline (0.66 and 0.37, respectively). The application of oversampling and focal loss proved effective in mitigating class imbalance. Beyond technical performance, the findings underline the practical utility of accurate sentiment detection for e-commerce platforms. Effective identification of negative feedback enables companies to address customer dis- satisfaction, refine product offerings, and inform personalized marketing strategies. This research contributes to advancing multilingual NLP in low-resource settings and provides a scalable solution for real-world sentiment classification in morphologically rich languages. The scientific novelty of this research lies in the integration of multilingual transformer architectures with adaptive classifiers and class imbalance mitigation techniques, specifically tailored for real-world e-commerce review data in morphologically rich languages. The proposed framework demonstrates how domain-specific fine-tuning and architectural customization can significantly enhance sentiment classification performance in low-resource language settings.
- 1 January 2021
- Published by Publishing House Helvetica (Publications)
- No. 176,p. 141-145
METHODOLOGY OF ANALYSIS OF THE CONTENT OF ELECTRONIC MEDIA
In the conditions of modern information society, the analysis of the flow of media information is of particular importance. Therefore, the issues of content analysis, which is based on the conversion of qualitative data into digital format in order to obtain information processing capabilities using technologies of big-data analytics and artificial intelligence approaches, are especially relevant. This study summarized the views on the use of content analysis and proposed a methodology for analyzing the content of electronic media for targeting and determining preferences for the content of publications. For this purpose, were proposed were proposed a conceptual scheme for conducting content analysis, which consists of 5 stages. The first stage consists of creating and updating a full-text database on a daily basis by automated data collection and retrieval of content from media websites (web scraping). The second stage involves using of modules for determining useful content (parsing) and its subsequent storage in the database. The quantitative analysis stage is especially important when generalization, aggregation of results, or to calculate the frequency of words (frequency analysis), categories, tags and expressions, as well as their joint use in the text. The purpose of qualitative analysis is to reduce the material so that the main content remains. Similarly, qualitative analysis can provide additional information about individual components of the text (terms, definitions) in order to improve understanding, explanation, and interpretation of a particular passage. Finally, the last stage is to analyze the tone of the texts (sentiment analysis), i.e. to determine the emotional (positive or negative) attitude of the authors of the test to the described event, person or phenomenon. Authors (or users who leave comments) actively use various means in order to create the desired picture of an event in the reader’s imagination, and to evoke the appropriate emotions. The proposed methodology is the basis for the implementation of modern tools of content analysis using information technologies and methods artificial intelligence approaches.
- 1 January 2021
- Published by Publishing House Helvetica (Publications)
- No. 176,p. 141-145
METHODOLOGY OF ANALYSIS OF THE CONTENT OF ELECTRONIC MEDIA
In the conditions of modern information society, the analysis of the flow of media information is of particular importance. Therefore, the issues of content analysis, which is based on the conversion of qualitative data into digital format in order to obtain information processing capabilities using technologies of big-data analytics and artificial intelligence approaches, are especially relevant. This study summarized the views on the use of content analysis and proposed a methodology for analyzing the content of electronic media for targeting and determining preferences for the content of publications. For this purpose, were proposed were proposed a conceptual scheme for conducting content analysis, which consists of 5 stages. The first stage consists of creating and updating a full-text database on a daily basis by automated data collection and retrieval of content from media websites (web scraping). The second stage involves using of modules for determining useful content (parsing) and its subsequent storage in the database. The quantitative analysis stage is especially important when generalization, aggregation of results, or to calculate the frequency of words (frequency analysis), categories, tags and expressions, as well as their joint use in the text. The purpose of qualitative analysis is to reduce the material so that the main content remains. Similarly, qualitative analysis can provide additional information about individual components of the text (terms, definitions) in order to improve understanding, explanation, and interpretation of a particular passage. Finally, the last stage is to analyze the tone of the texts (sentiment analysis), i.e. to determine the emotional (positive or negative) attitude of the authors of the test to the described event, person or phenomenon. Authors (or users who leave comments) actively use various means in order to create the desired picture of an event in the reader’s imagination, and to evoke the appropriate emotions. The proposed methodology is the basis for the implementation of modern tools of content analysis using information technologies and methods artificial intelligence approaches.
- 11 November 2020
- Published by Kyiv National Economic University named after Vadym Hetman in Neuro-Fuzzy Modeling Techniques in Economics
- No. 9,p. 95-137
Machine learning approach of analysis of emotional polarity of electronic social media
У статті пропонується новий підхід до оцінки емоційної полярності (або аналізу настроїв) електронних текстів у соціальних мережах. Для цього використовувалися як класичні методи машинного навчання (логістична регресія та метод опорних векторів), так і інструментарій глибоких нейронних мереж (повнозв’язні та згорткові нейромережі). Векторне представлення ґрунтувалось на частотних та попередньо навчених вкладеннях слів Word2vec і GloVe (з розмірами вкладення 100 і 300). Для вибраного англомовного набору даних IMDb Movie Reviews точність класифікації за допомогою моделі логістичної регресії становила 87%, машини опорних векторів – 87,5%, повнозв’язної нейронної мережі – 88% і згорткової мережі – 90%. Точність запропонованих моделей є цілком прийнятною для практичних ситуацій і не поступається передовим рішенням у сфері обробки природньої мови за напрямом аналізу настроїв, що відкриває обнадійливі перспективи для подальших досліджень This paper proposes a new approach to evaluating the emotional polarity (or Sentiment Analysis) of electronic social media texts. For this purpose both conventional Machine Learning (Logistic Regression and Support Vector Machine), and Deep Neural Networks approaches (Fully Connected and Convolutional Neural Networks) were used. As vector representations of words, we used both the frequency-based and pretrained words embeddings Word2vec and GloVe (with embedding dimensions of size 100 and 300). For the selected English-language IMDb Movie Reviews dataset the classification accuracy using the Logistic Regression model was 87%, the Support Vector Machine – 87.5%, the Fully Connected Neural Network – 88%, and the Convolutional Network – 90%. The accuracy of the proposed models is a quite acceptable for practical use-cases and is not inferior to cutting-edge Natural Language Processing solutions in the field of Sentiment Analysis, which opens up good prospects for further research.
- 11 November 2020
- Published by Kyiv National Economic University named after Vadym Hetman in Neuro-Fuzzy Modeling Techniques in Economics
- No. 9,p. 95-137
Machine learning approach of analysis of emotional polarity of electronic social media
У статті пропонується новий підхід до оцінки емоційної полярності (або аналізу настроїв) електронних текстів у соціальних мережах. Для цього використовувалися як класичні методи машинного навчання (логістична регресія та метод опорних векторів), так і інструментарій глибоких нейронних мереж (повнозв’язні та згорткові нейромережі). Векторне представлення ґрунтувалось на частотних та попередньо навчених вкладеннях слів Word2vec і GloVe (з розмірами вкладення 100 і 300). Для вибраного англомовного набору даних IMDb Movie Reviews точність класифікації за допомогою моделі логістичної регресії становила 87%, машини опорних векторів – 87,5%, повнозв’язної нейронної мережі – 88% і згорткової мережі – 90%. Точність запропонованих моделей є цілком прийнятною для практичних ситуацій і не поступається передовим рішенням у сфері обробки природньої мови за напрямом аналізу настроїв, що відкриває обнадійливі перспективи для подальших досліджень This paper proposes a new approach to evaluating the emotional polarity (or Sentiment Analysis) of electronic social media texts. For this purpose both conventional Machine Learning (Logistic Regression and Support Vector Machine), and Deep Neural Networks approaches (Fully Connected and Convolutional Neural Networks) were used. As vector representations of words, we used both the frequency-based and pretrained words embeddings Word2vec and GloVe (with embedding dimensions of size 100 and 300). For the selected English-language IMDb Movie Reviews dataset the classification accuracy using the Logistic Regression model was 87%, the Support Vector Machine – 87.5%, the Fully Connected Neural Network – 88%, and the Convolutional Network – 90%. The accuracy of the proposed models is a quite acceptable for practical use-cases and is not inferior to cutting-edge Natural Language Processing solutions in the field of Sentiment Analysis, which opens up good prospects for further research.