Publications

@inproceedings{yeshpanov-etal-2024-kazqad-kazakh, title = "{K}az{QAD}: {K}azakh Open-Domain Question Answering Dataset", author = "Yeshpanov, Rustem and Efimov, Pavel and Boytsov, Leonid and Shalkarbayuli, Ardak and Braslavski, Pavel", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.843", pages = "9645--9656", abstract = "We introduce KazQAD{---}a Kazakh open-domain question answering (ODQA) dataset{---}that can be used in both reading comprehension and full ODQA settings, as well as for information retrieval experiments. KazQAD contains just under 6,000 unique questions with extracted short answers and nearly 12,000 passage-level relevance judgements. We use a combination of machine translation, Wikipedia search, and in-house manual annotation to ensure annotation efficiency and data quality. The questions come from two sources: translated items from the Natural Questions (NQ) dataset (only for training) and the original Kazakh Unified National Testing (UNT) exam (for development and testing). The accompanying text corpus contains more than 800,000 passages from the Kazakh Wikipedia. As a supplementary dataset, we release around 61,000 question-passage-answer triples from the NQ dataset that have been machine-translated into Kazakh. We develop baseline retrievers and readers that achieve reasonable scores in retrieval (NDCG10 = 0.389 MRR = 0.382), reading comprehension (EM = 38.5 F1 = 54.2), and full ODQA (EM = 17.8 F1 = 28.7) settings. Nevertheless, these results are substantially lower than state-of-the-art results for English QA collections, and we think that there should still be ample room for improvement. We also show that the current OpenAI{'}s ChatGPTv3.5 is not able to answer KazQAD test questions in the closed-book setting with acceptable quality. The dataset is freely available under the Creative Commons licence (CC BY-SA) at url https://github.com/IS2AI/KazQAD", }

@ARTICLE{10509679, author={Keutayeva, Aigerim and Abibullaev, Berdakh}, journal={IEEE Access}, title={Data Constraints and Performance Optimization for Transformer-Based Models in EEG-Based Brain-Computer Interfaces: A Survey}, year={2024}, volume={12}, number={}, pages={62628-62647}, keywords={Brain modeling;Electroencephalography;Transformers;Data models;Decoding;Task analysis;Training;Brain-computer interfaces;data scarcity;survey;deep learning;transformer;EEG;self-attention mechanism;motor imagery;BCI}, doi={10.1109/ACCESS.2024.3394696}}

@inproceedings{maxutov-etal-2024-llms, title = "Do {LLM}s Speak {K}azakh? A Pilot Evaluation of Seven Models", author = "Maxutov, Akylbek and Myrzakhmet, Ayan and Braslavski, Pavel", editor = {Ataman, Duygu and Derin, Mehmet Oguz and Ivanova, Sardana and K{\"o}ksal, Abdullatif and S{\"a}lev{\"a}, Jonne and Zeyrek, Deniz}, booktitle = "Proceedings of the First Workshop on Natural Language Processing for Turkic Languages (SIGTURK 2024)", month = aug, year = "2024", address = "Bangkok, Thailand and Online", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2024.sigturk-1.8", pages = "81--91", abstract = "We conducted a systematic evaluation of seven large language models (LLMs) on tasks in Kazakh, a Turkic language spoken by approximately 13 million native speakers in Kazakhstan and abroad. We used six datasets corresponding to different tasks {--} questions answering, causal reasoning, middle school math problems, machine translation, and spelling correction. Three of the datasets were prepared for this study. As expected, the quality of the LLMs on the Kazakh tasks is lower than on the parallel English tasks. GPT-4 shows the best results, followed by Gemini and . In general, LLMs perform better on classification tasks and struggle with generative tasks. Our results provide valuable insights into the applicability of currently available LLMs for Kazakh. We made the data collected for this study publicly available: https://github.com/akylbekmaxutov/LLM-eval-using-Kazakh.", }

@Article{buildings14040935, AUTHOR = {Karaca, Ferhat and Tleuken, Aidana and Pineda-Martos, Rocío and Ros Cardoso, Sara and Orel, Daniil and Askar, Rand and Agibayeva, Akmaral and Güemez, Elena Goicolea and Salles, Adriana and Varol, Huseyin Atakan and Braganca, Luis}, TITLE = {Cultivating Sustainable Construction: Stakeholder Insights Driving Circular Economy Innovation for Inclusive Resource Equity}, JOURNAL = {Buildings}, VOLUME = {14}, YEAR = {2024}, NUMBER = {4}, ARTICLE-NUMBER = {935}, URL = {https://www.mdpi.com/2075-5309/14/4/935}, ISSN = {2075-5309}, ABSTRACT = {Due to its intricate production processes, complex supply chains, and industry-specific characteristics, the construction industry faces unique challenges in adopting circular economy (CE) principles that promote resource equity. To address this issue, this study aims to delve into identifying stakeholders’ opinions and perceptions regarding key CE strategies across different stages of the building life cycle (BLC). Both European and non-European stakeholders within the “CircularB” COST Action network and beyond participated in this research. Three methods were employed to assess stakeholders’ opinions: an online survey, a structured survey with a semi-guided workshop, and creative thinking round table discussions. Natural language processing (NLP), specifically topic modelling and sentiment analysis, was used to analyse the data collected from the online survey, which gathered text-based opinions from 209 participants on the cost-benefit aspects of circularity strategies. The structured survey, which collected data from 43 workshop participants, evaluated the perceived importance of CE strategies across various BLC phases and assessed the adoption of selected CE strategies in current or past projects. Finally, the Six Thinking Hats® activity, employed in the round table discussions, generated ideas from 25 professionals regarding the broader implementation challenges and opportunities of CE in construction. The research findings highlight the need to bridge the gap between theory and practice by fostering active industry stakeholder involvement in the transition to a CE model. The analyses of the collected stakeholder opinions through the three activities contribute to proactive and collaborative efforts aimed at advancing resource equity in the construction sector and promoting just and inclusive resource use. In summary, this research offers a comprehensive understanding of stakeholders’ opinions on CE strategies and provides guidance for the development of targeted policies and strategies to accelerate the integration of CE principles in the construction industry.}, DOI = {10.3390/buildings14040935} }

@article{https://doi.org/10.1002/eem2.12808, author = {Mubarak, Assem and Sarsembayev, Bayandy and Serik, Yerzhigit and Onabek, Abdirakhman and Kappassov, Zhanat and Bakenov, Zhumabay and Tsuchiya, Kazuyoshi and Kalimuldina, Gulnur}, title = {Quenched PVDF/PMMA Porous Matrix for Triboelectric Energy Harvesting and Sensing}, journal = {ENERGY \& ENVIRONMENTAL MATERIALS}, volume = {n/a}, number = {n/a}, pages = {e12808}, keywords = {N,N′-poly(methyl methacrylate) (PMMA), robotic tactile sensors, polyvinylidene fluoride (PVDF), quenching, triboelectric nanogenerators (TENGs)}, doi = {https://doi.org/10.1002/eem2.12808}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1002/eem2.12808}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1002/eem2.12808}, note = {e12808 EEM-2024-0184.R1}, abstract = {The rapid development of nanotechnology has significantly revolutionized wearable electronics and expanded their functionality. Through introducing innovative solutions for energy harvesting and autonomous sensing, this research presents a cost-effective strategy to enhance the performance of triboelectric nanogenerators (TENGs). The TENG was fabricated from polyvinylidene fluoride (PVDF) and N, N′-poly(methyl methacrylate) (PMMA) blend with a porous structure via a novel optimized quenching method. The developed approach results in a high β-phase content (85.7\%) PVDF/3wt.\%PMMA porous blend, known for its superior piezoelectric properties. PVDF/3wt.\%PMMA modified porous TENG demonstrates remarkable electrical output, with a dielectric constant of 40 and an open-circuit voltage of approximately 600 V. The porous matrix notably increases durability, enduring over 36 000 operational cycles without performance degradation. Moreover, practical applications were explored in this research, including powering LEDs and pacemakers with a maximum power output of 750 mW m−2. Also, TENG served as a self-powered tactile sensor for robotic applications in various temperature conditions. The work highlights the potential of the PVDF/PMMA porous blend to utilize the next-generation self-powered sensors and power small electronic devices.}}

@article{https://doi.org/10.1111/jiec.13502, author = {Tleuken, Aidana and Orel, Daniil and Iskakova, Anel and Varol, Huseyin Atakan and Karaca, Ferhat}, title = {Exploring stakeholders’ opinions on circular economy in the construction sector: A natural language processing analysis of social media discourse}, journal = {Journal of Industrial Ecology}, volume = {n/a}, number = {n/a}, pages = {}, keywords = {circular economy, construction sector, industrial ecology, sentiment analysis, sustainable buildings, topic modeling}, doi = {https://doi.org/10.1111/jiec.13502}, url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/jiec.13502}, eprint = {https://onlinelibrary.wiley.com/doi/pdf/10.1111/jiec.13502}, abstract = {Abstract The construction industry has been criticized for its negative environmental impacts, leading industry experts to advocate for a shift toward a circular economy (CE) model. However, there is a lack of research on stakeholder opinions regarding that. This research paper examines stakeholders’ perspectives on implementing CE principles in the construction industry by conducting artificial intelligence-powered natural language processing (NLP) through online sources. It answers three questions: What themes and concepts are associated with the CE in construction? How do opinions on the CE vary across different online platforms? And what factors shape positive attitudes toward the CE? The data obtained from various platforms showed that 57\% of sentiments were positive, 28\% were neutral, and 15\% were negative. This research provides critical knowledge on the analysis of CE representation on social media in construction. Moreover, a webpage tool has been created that can assess any input opinion on the scale (positive, neutral, or negative) for further use (https://ce-sentiment.streamlit.app/). This NLP-based research of social media discourse in the construction sector can directly influence policy decisions by offering real-time insights into public sentiment and preferences, shaping regulations that align with societal needs. It also provides industry professionals with data-driven guidance, enabling them to identify growth opportunities and innovation pathways within the CE, ultimately fostering a more sustainable and prosperous future.} }

@InProceedings{10.1007/978-3-031-57800-7_31, author="Karaca, Ferhat and Tleuken, Aidana and Pineda-Martos, Roc{\'i}o and Cardoso, Sara Ros and Askar, Rand and Salles, Adriana and G{\"u}emez, Elena Goicolea and Agibayeva, Akmaral and Varol, Huseyin Atakan and Braganca, Luis", editor="Ungureanu, Viorel and Bragan{\c{c}}a, Lu{\'i}s and Baniotopoulos, Charalambos and Abdalla, Khairedin M.", title="Analysing Stakeholder Opinions Within the COST Action CA21103 CircularB and Beyond: Circular Economy Implementation in Construction", booktitle="4th International Conference ``Coordinating Engineering for Sustainability and Resilience'' {\&} Midterm Conference of CircularB ``Implementation of Circular Economy in the Built Environment''", year="2024", publisher="Springer Nature Switzerland", address="Cham", pages="335--345", abstract="This study examines the importance and adoption of circular economy (CE) principles within the construction industry by focusing on stakeholders' opinions on key CE strategies across different building life cycle stages. The study draws insights from the perceptions of European-based stakeholders who actively participated in the CircularB Workshop 1 Part 2, entitled Creating a Roadmap towards Circularity in the Built Environment - State-of-the-Art. The research comprises two parts. In the first part, a structured survey was employed to systematically collect opinions on the levels of awareness and variations among the adoption and importance of selected CE implementation strategies within the construction sector. The second part engaged stakeholders in a dynamic creative thinking activity, posing seven targeted questions allowing participants to offer multiple answers for each query. Overall, the study sheds light on the multifaceted challenges and opportunities inherent in fostering CE within the construction domain by highlighting the significance of recognising and addressing systemic barriers within the CE framework, the importance of product design for disassembly, and the efficient production of reusable and recyclable materials. Furthermore, it emphasises the necessity to motivate industry stakeholders to participate actively in the transition to a CE, bridging the gap between theory and practice frameworks and increasing the engagement of policymakers and governments.", isbn="978-3-031-57800-7" }

@ARTICLE{10595079, author={Abdrakhmanova, Madina and Yermekova, Assel and Barko, Yuliya and Ryspayev, Vladislav and Jumadildayev, Medet and Varol, Huseyin Atakan}, journal={IEEE Access}, title={One Model to Rule Them all: A Universal Transformer for Biometric Matching}, year={2024}, volume={12}, number={}, pages={96729-96739}, keywords={Transformers;Vectors;Feature extraction;Visualization;Speech recognition;Biological system modeling;Task analysis;Biometrics (access control);Biometric matching;cross-modal matching;face verification;face-audio association;metric learning;multimodal verification;speaker verification;transformer}, doi={10.1109/ACCESS.2024.3426602}}

@INPROCEEDINGS{10581992, author={Kuzdeuov, Askat and Taratynova, Darya and Tleuliyev, Alim and Varol, Huseyin Atakan}, booktitle={2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG)}, title={OpenThermalPose: An Open-Source Annotated Thermal Human Pose Dataset and Initial YOLOv8-Pose Baselines}, year={2024}, volume={}, number={}, pages={1-8}, keywords={Privacy;Annotations;Source coding;Pose estimation;Lighting;Medical services;Motion capture}, doi={10.1109/FG59268.2024.10581992}}

@inproceedings{yeshpanov-etal-2024-kazparc-kazakh, title = "{K}az{P}ar{C}: {K}azakh Parallel Corpus for Machine Translation", author = "Yeshpanov, Rustem and Polonskaya, Alina and Varol, Huseyin Atakan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.842", pages = "9633--9644", abstract = "We introduce KazParC, a parallel corpus designed for machine translation across Kazakh, English, Russian, and Turkish. The first and largest publicly available corpus of its kind, KazParC contains a collection of 371,902 parallel sentences covering different domains and developed with the assistance of human translators. Our research efforts also extend to the development of a neural machine translation model nicknamed Tilmash. Remarkably, the performance of Tilmash is on par with, and in certain instances, surpasses that of industry giants, such as Google Translate and Yandex Translate, as measured by standard evaluation metrics such as BLEU and chrF. Both KazParC and Tilmash are openly available for download under the Creative Commons Attribution 4.0 International License (CC BY 4.0) through our GitHub repository.", }

@inproceedings{abilbekov-etal-2024-kazemotts-dataset, title = "{K}az{E}mo{TTS}: A Dataset for {K}azakh Emotional Text-to-Speech Synthesis", author = "Abilbekov, Adal and Mussakhojayeva, Saida and Yeshpanov, Rustem and Varol, Huseyin Atakan", editor = "Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.841", pages = "9626--9632", abstract = "This study focuses on the creation of the KazEmoTTS dataset, designed for emotional Kazakh text-to-speech (TTS) applications. KazEmoTTS is a collection of 54,760 audio-text pairs, with a total duration of 74.85 hours, featuring 34.23 hours delivered by a female narrator and 40.62 hours by two male narrators. The list of the emotions considered include {`}neutral{''}, {}angry{''}, {}happy{''}, {}sad{''}, {}scared{''}, and {`}surprised{''}. We also developed a TTS model trained on the KazEmoTTS dataset. Objective and subjective evaluations were employed to assess the quality of synthesized speech, yielding an MCD score within the range of 6.02 to 7.67, alongside a MOS that spanned from 3.51 to 3.57. To facilitate reproducibility and inspire further research, we have made our code, pre-trained model, and dataset accessible in our GitHub repository.", }