基于俄汉新闻网评可比语料库的情感分析研究

Sentimental Analysis Based on Russian-Chinese Comparable Corpus of News Comments

本研究自建小型俄汉新闻网络评论可比语料库,设计了情感标注体系,对同一新闻主题的30 426条俄汉网评中表达情感的名词、动词、形容词、副词和词组进行了标注,以进行情感倾向性判定和表达手段对比。结果发现:构建的可比语料库和标注体系可有效用于情感倾向性半自动分析及对比中,要提高分析精度还需考虑其他词类、修辞手法、标点符号等对情感表达的影响,而情感词典可借助语料标注搭建;中文网评趋向使用短评论、动词和名词表达情感,而俄文网络评论趋向使用长评论、形容词和动词表达情感;俄汉网评中均存在情感评价主题迁移现象,中文此类现象略多。该成果可为俄语话语分析、情感识别、舆情监控等研究提供帮助。

In this study,a small Russian-Chinese comparable corpus of news comments was compiled and a sentimental annotation system was designed.The nouns,verbs,adjectives,adverbs and collocations expressing emotions in 30 426 Russian-Chinese online comments of the same news subject were annotated in order to compare emotional tendency and means of sentimental expression.The results show that the comparable corpus and the annotation system can be used in semi-automatic analysis and comparison of sentimental tendencies;in order to improve the accuracy,the influence of other parts of speech,rhetorical devices and punctuation on emotional expression should be considered;Chinese reviews tend to use short comments,nouns and verbs to express emotions,while Russian online reviews tend to use long comments,adjectives and verbs to express emotions.The results can be helpful for research on Russian discourse analysis,emotion recognition,and public opinion monitoring.