网络评论文本情感极性的语言学研究

发布时间：2019-01-10 18:00

【摘要】：文本情感倾向性研究属于计算语言学、人工智能、数据挖掘等不同学科的交叉领域，但是这几年的研究主要偏重技术路线，比如：主观性句子的识别技术，利用词汇和语义判别观点句子情感极性的技术，而从语言学本体的角度进行文本情感倾向分析的文献较少。文本情感倾向分析不但涉及情感信息的挖掘，而且还关系到情感表达的语言知识发现，比如情感表达是否有明确的语言学标记，情感表达是否有较为偏爱的句类、句式或者词类、语气，情感表达的语言学现象能否用已有的语言学理论来阐释等等，简而言之就是，文本情感倾向分析能否找到情感表达的语言学方程式。本研究以网络评论文本（针对手机的评论）为语料，对其蕴含明确情感倾向的句子进行词类、句型、特定结构、语义、语用等角度的深入分析，，以语言的主观性和主观化研究成果为基础，从语义标注、语用分析等角度归纳出正、负情感倾向的语言学标识。本文共分五部分。第一章绪论，介绍了本研究的选题背景、研究目的、方法、内容和研究现状，并对语料来源做了简要说明。第二章理论基础，从文本的主观性、主观化、语气、情态等角度探讨了文本倾向性分析的可能性。第三章、第四章分别从词语层面和句子层面对文本的情感倾向性做了分析，重点探讨了语气副词、否定副词、程度副词、疑问代词、指示代词以及部分特殊句式在文本情感倾向性分析中的作用和价值，总结出127个包含上述成分的情感表达式，并用实际语料中的例句进行说明。第五章结论，总结了全文的工作，讨论了从语言学角度进行文本倾向性分析的意义和可能的成果，并对下一步的工作进行展望。通过对语料的考察，我们发现表达负面情感的主要手段是使用否定结构，否定结构和语气副词、程度副词、指示代词、疑问代词的连用加强或削弱了否定的效果，从而表达出不同强度的负面情感。表达正面情感的主要手段是使用褒义词或贬义词的否定形式，词的褒贬性很多时候是受具体语境影响的，正面情感的表达式需要比负面情感的表达式有更多的上下文知识来说明。本文通过建立手机评论语料的情感词表，力求达到对词语或句式情感倾向性的细致描写。
[Abstract]:The research on text affective orientation belongs to the intersecting fields of computational linguistics, artificial intelligence, data mining and so on. However, in recent years, the research has mainly focused on the technical route, such as: subjective sentence recognition technology. Lexical and semantic techniques are used to judge the affective polarity of sentences, but there is less literature to analyze the affective tendency of the text from the perspective of linguistic ontology. Text affective tendency analysis not only involves the mining of emotional information, but also relates to language knowledge discovery of emotional expression, such as whether emotional expression has clear linguistic markers and whether emotional expression has a preferred sentence category. Whether the linguistic phenomena of sentence structure or part of speech, mood, emotional expression can be explained by existing linguistic theories, in short, can text affective tendency analysis find the linguistic equation of emotional expression. In this study, the text of online comments (reviews for mobile phones) is used as the corpus to make a thorough analysis of the parts of speech, sentence patterns, specific structures, semantics, pragmatics, and so on, which contain clear emotional tendencies. Based on the research results of subjectivity and subjectivization of language, the linguistic markers of positive and negative affective tendencies are summed up from the perspectives of semantic tagging and pragmatic analysis. This paper is divided into five parts. The first chapter introduces the background, purpose, method, content and current situation of the research, and gives a brief description of the source of the corpus. The second chapter discusses the possibility of text orientation analysis from the aspects of subjectivity, mood, modality and so on. Chapter three, Chapter four analyzes the emotional tendency of the text from the word level and sentence level, focusing on mood adverbs, negative adverbs, degree adverbs, interrogative pronouns. The role and value of demonstrative pronouns and some special sentence patterns in the analysis of the affective tendency of the text are summarized. 127 emotional expressions containing the above components are summarized and illustrated by the example sentences in the actual corpus. The fifth chapter summarizes the work of the paper, discusses the significance and possible achievements of text orientation analysis from the linguistic point of view, and looks forward to the next work. Through the investigation of the corpus, we find that the main means of expressing negative emotions are to use negative structure, negative structure and mood adverb, degree adverb, demonstrative pronoun and interrogative pronoun to strengthen or weaken the negative effect. Thus expressing different intensity of negative emotions. The main means of expressing positive emotion is to use the negative form of positive or derogatory words. The praise and derogation of words are often influenced by specific context. Expressions of positive emotions need more contextual knowledge than expressions of negative emotions. In this paper, the emotional lexicon of mobile phone commentary corpus is established to describe the sentiment tendency of words or sentence patterns in detail.
【学位授予单位】：河北大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：H146

【参考文献】