Construction and Research of Multilingual Parallel Emotion Corpus

doi:10.3778/j.issn.1673-9418.2412006

Abstract

Abstract: In the field of natural language processing, the construction and application of multilingual parallel corpora has increasingly become a research hotspot, and its application value in the fields of cross-cultural communication, machine translation, sentiment analysis, and intelligent question answering has become more and more prominent. Against the backdrop of “the Belt and Road Initiative”, it is of great significance to deepen exchanges and cooperation between China and Mongolia in the cultural, economic and social fields. Therefore, this paper constructs a multilingual parallel emotion corpus covering Chinese, traditional Mongolian and cyrillic Mongolian to promote sentiment analysis research in the Mongolian system. The construction of this corpus will not only help deepen the cultural exchanges and cooperation between China and Mongolia, but also promote the unity of Chinese nations and the inheritance and development of Mongolian ethnic culture. Based on the two public Chinese sentiment analysis datasets of SMP2020-EWECT and OCEMOTION, this paper first uses advanced machine translation technology to automatically translate Chinese texts into traditional Mongolian and cyrillic Mongolian texts, and ensures the accuracy and high quality of the corpus through a dual verification mechanism combining machine proofreading and human proofreading. Then, detailed data analysis is carried out on the corpus to ensure the diversity and balance of the text, covering different domains and sentiment categories, and ensuring the wide applicability and representativeness of the corpus in a variety of natural language processing tasks. Finally, through a series of systematic experiments, the performance of various sentiment analysis models on Mongolian data is explored, and the experimental results verify the usability and research value of the corpus in sentiment analysis tasks.

Key words: natural language processing, multilingual parallel corpus, traditional Mongolian, cyrillic Mongolian

摘要： 在自然语言处理领域，多语平行语料库的构建与应用日益成为研究热点，其在跨文化交流、机器翻译、情感分析及智慧问答等领域的应用价值愈发凸显。特别是在“一带一路”倡议背景下，深化中蒙两国在文化、经济及社会领域的交流合作具有重要意义。因此，该研究构建了一个涵盖中文、传统蒙文和新蒙文的多语平行情感语料库，以推动蒙文体系下的情感分析研究。该语料库的构建不仅有助于深化中蒙两国间的文化交流与合作，还将促进我国民族团结及蒙古族文化的传承与发展。该研究基于SMP2020-EWECT和OCEMOTION两个公开中文情感分析数据集，采用先进的机器翻译技术，将中文文本自动翻译为传统蒙文和新蒙文，并通过机器校对与人工校对相结合的双重验证机制，确保语料库的准确性和高质量。对该语料库进行详尽的数据分析，以确保文本的多样性与平衡性，涵盖不同领域和情感类别，保证语料库在多种自然语言处理任务中的广泛适用性和代表性。通过一系列系统性的实验，探索了多种情感分析模型在蒙文数据上的性能表现，实验结果验证了该语料库在情感分析任务中的可用性和研究价值。

关键词: 自然语言处理, 多语言平行语料库, 传统蒙文, 新蒙文

ZHANG Hao, LI Shiqi, DIAO Yufeng, YANG Liang, LIN Hongfei, FAN Xiaochao. Construction and Research of Multilingual Parallel Emotion Corpus[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(9): 2506-2519.

张浩, 李世琪, 刁宇峰, 杨亮, 林鸿飞, 樊小超. 多语言平行情感语料库的构建与研究[J]. 计算机科学与探索, 2025, 19(9): 2506-2519.

References

[1] 清格尔泰. 蒙古语语法[M]. 呼和浩特: 内蒙古人民出版社, 1991.
CHINGGELTEI. Mongolian grammar[M]. Hohhot: Inner Mongolia People??s Publishing House, 1991.
[2] JIANG Y E, LIU T Y, MA S M, et al. Discourse-centric evaluation of document-level machine translation with a new densely annotated parallel corpus of novels[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2023: 7853-7872.
[3] LEE Y, OH P, THORNE J. Knowledge corpus error in question answering[C]//Findings of the Association for Computational Linguistics: EMNLP 2023. Stroudsburg: ACL, 2023: 9183-9197.
[4] GOKANI M, MAMIDI R. GSAC: a Gujarati sentiment analysis corpus from Twitter[C]//Proceedings of the 13th Workshop on Computational Approaches to Subjectivity, Sentiment & Social Media Analysis. Stroudsburg: ACL, 2023: 129-137.
[5] PANG B, LEE L, VAITHYANATHAN S. Thumbs up? : sentiment classification using machine learning techniques[C]//Proceedings of the 2002 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2002: 79-86.
[6] HU M Q, LIU B. Mining and summarizing customer reviews[C]//Proceedings of the 10th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2004: 168-177.
[7] 姚源林, 王树伟, 徐睿峰, 等. 面向微博文本的情绪标注语料库构建[J]. 中文信息学报, 2014, 28(5): 83-91.
YAO Y L, WANG S W, XU R F, et al. The construction of an emotion annotated corpus on microblog text[J]. Journal of Chinese Information Processing, 2014, 28(5): 83-91.
[8] 徐琳宏, 林鸿飞, 赵晶. 情感语料库的构建和分析[J]. 中文信息学报, 2008, 22(1): 116-122.
XU L H, LIN H F, ZHAO J. Construction and analysis of emotional corpus[J]. Journal of Chinese Information Processing, 2008, 22(1): 116-122.
[9] 荀恩东, 饶高琦, 肖晓悦, 等. 大数据背景下BCC语料库的研制[J]. 语料库语言学, 2016, 3(1): 93-109.
XUN E D, RAO G Q, XIAO X Y, et al. The construction of the BCC corpus in the age of big data[J]. Corpus Linguistics, 2016, 3(1): 93-109.
[10] 黄水清, 王东波. 国内语料库研究综述[J]. 信息资源管理学报, 2021, 11(3): 4-17.
HUANG S Q, WANG D B. Review of corpus research in China[J]. Journal of Information Resources Management, 2021, 11(3): 4-17.
[11] FRANCIS W N. A standard corpus of edited present-day American English[J]. College English, 1965, 26(4): 267-273.
[12] LEECH G. 100 million words of English: the British national corpus (BNC)[J]. Second Language Research, 1992, 28: 1-13.
[13] DAVIES M. The corpus of contemporary American English as the first reliable monitor corpus of English[J]. Literary and Linguistic Computing, 2010, 25(4): 447-464.
[14] 薛松. 汉英平行语料库中名词短语对齐算法的研究[D]. 北京: 中国科学院研究生院(软件研究所), 2003.
XUE S. Research on noun phrase alignment algorithm in Chinese-English parallel corpus[D]. Beijing: Graduate School of the Chinese Academy of Sciences (Institute of Software), 2003.
[15] LI Y Y, LUO Y F, LIN Y, et al. A simple and effective approach to robust unsupervised bilingual dictionary induction[C]//Proceedings of the 28th International Conference on Computational Linguistics, 2020: 5990-6001.
[16] BROWN P F, LAI J C, MERCER R L. Aligning sentences in parallel corpora[C]//Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 1991: 169-176.
[17] 柏晓静, 常宝宝, 詹卫东, 等. 构建大规模的汉英双语平行语料库[C]//机器翻译研究进展-2002年全国机器翻译研讨会论文集, 2002: 131-138.
BAI X J, CHANG B B, ZHAN W D, et al. Construction of a large-scale Chinese-English bilingual parallel corpus[C]//Progress in Machine Translation Research-2002 National Machine Translation Symposium, 2002: 131-138.
[18] 嘎日迪, 斯日古楞. 蒙古文信息处理技术及自然语言理解[M]. 呼和浩特: 内蒙古大学出版社, 2006.
GARDI, SERGULENG. Mongolian information processing technology and natural language understanding[M]. Hohhot: Inner Mongolia University Press, 2006.
[19] 王斯日古楞, 德·萨日娜, 那顺乌日图. 现代蒙古语谓语段自动标注系统的设计与实现[C]//第十一届全国民族语言文字信息学术研讨会论文集, 2007: 430-437.
WANG SERGULENG, D·SARNA, NASNURT. Design and implementation of an automatic tagging system for predicate segments in modern Mongolian[C]//Proceedings of the 11th National Symposium on Ethnic Language and Character Information, 2007: 430-437.
[20] 斯日古楞. 《现代蒙古语语料库管理平台》建设[D]. 呼和浩特: 内蒙古大学, 2010.
SERGULENG. Construction of “modern Mongolian corpus management platform”[D]. Hohhot: Inner Mongolia University, 2010.
[21] 淑琴, 那顺乌日图. 面向EBMT系统的汉蒙双语语料库的构建[J]. 内蒙古社会科学, 2006, 27(1): 140-144.
SHU Q, Nasnurt. Construction of a Chinese-Mongolian bilingual corpus for EBMT system[J]. Inner Mongolia Social Sciences, 2006, 27(1): 140-144.
[22] RYKIN P. Mongolic historical and comparative linguistics: state-of-the-art and recent advances[J]. Central Asiatic Journal, 2021, 64(1/2).
[23] BATBAYAR T. Geopolitics and Mongolia??s search for Post-Soviet identity[J]. Eurasian Geography and Economics, 2002, 43(4): 323-335.
[24] 包萨日娜. 传统蒙古文到新蒙文转换中名词及其格附加成分转换的研究[D]. 呼和浩特: 内蒙古大学, 2009.
BAO SARNA. Research on the conversion of nouns and their case affixes in the conversion from traditional Mongolian to new Mongolian[D]. Hohhot: Inner Mongolia University, 2009.
[25] 包月英. 蒙古文单词个性化学习系统的设计与实践研究[D]. 呼和浩特: 内蒙古师范大学, 2020.
BAO Y Y. Design and practical research of a personalized learning system for Mongolian words[D]. Hohhot: Inner Mongolia Normal University, 2020.
[26] 杨帆. 基于整词的蒙古文在线手写识别研究与实现[D]. 呼和浩特: 内蒙古大学, 2021.
YANG F. Research and implementation of online handwritten recognition of Mongolian based on whole words[D]. Hohhot: Inner Mongolia University, 2021.
[27] 格根. 蒙古文字史略及俄语字母在西里尔蒙古文中的表音问题[D]. 广州: 广东外语外贸大学, 2020.
GE G. A brief history of Mongolian writing and the pronunciation of Russian letters in cyrillic Mongolian[D]. Guangzhou: Guangdong University of Foreign Studies, 2020.
[28] RIFTIN B. Mongolian translations of old Chinese novels and stories-a tentative bibliographic survey[M]//Literary migrations: traditional Chinese fiction in Asia (17th-20th centuries). Singapore: ISEAS-Yusof Ishak Institute, 2013: 130-160.
[29] QU W S, LI R. Translation of personal and place names from and into Chinese in modern China: a lexicographical history perspective[J]. International Journal for the Semiotics of Law, 2015, 28(3): 525-557.
[30] KIM Y. Convolutional neural networks for sentence classification[C]//Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2014: 1746-1751.
[31] 曹宇, 李天瑞, 贾真, 等. BGRU: 中文文本情感分析的新方法[J]. 计算机科学与探索, 2019, 13(6): 973-981.
CAO Y, LI T R, JIA Z, et al. BGRU: new method of Chinese text sentiment analysis[J]. Journal of Frontiers of Computer Science and Technology, 2019, 13(6): 973-981.
[32] WANG Y Q, HUANG M L, ZHU X Y, et al. Attention-based LSTM for aspect-level sentiment classification[C]//Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing. Stroudsburg: ACL, 2016: 606-615.
[33] KUMAR A, KAWAHARA D, KUROHASHI S. Knowledge-enriched two-layered attention network for sentiment analysis[C]//Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2018: 253-258.
[34] XU H, LIU B, SHU L, et al. BERT post-training for review reading comprehension and aspect-based sentiment analysis[EB/OL]. [2024-10-18]. https://arxiv.org/abs/1904.02232.
[35] DEVLIN J, CHANG M W, LEE K, et al. BERT: pre-training of deep bidirectional transformers for language understanding[C]//Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Stroudsburg: ACL, 2019: 4171-4186.