Study On The Sentiment Polarity Of Network Reviews In The Perspective Of Linguistics

Text sentiment analysis is the intersection of different disciplines, such as computationallinguistics, artificial intelligence, and data mining. In recent years, the research mainlyfocuses on the technical route. For example, the subjective sentence recognition technology,the lexical and semantic analysis etc. But from the perspective of linguistics, research of textsentiment analysis is less mentioned. Text sentiment polarity not only involves the datamining, but also related to the discovery of emotional knowledge, such as whether there is aclear linguistic mark about the emotional expression; whether there are some types ofsentences, words, or mood that are preferred by the emotional expression; whether thesentiment analysis could be described by the present linguistic theories.This thesis uses the mobile phone review as corpus, based on the Chinese Subjectivityand Subjectivisation research, analyses the emotional sentences or phrases from the view ofwords, structure, semantics, and pragmatics in hope of discovering the positive and negativeemotional linguistic equation.The paper is divided into five parts. Chapter I introduces the research background, thepurpose, method, content and status. Besides, corpus sources are briefly described. Chapter IIis the theory. Form the point of subjectivity, subjectivisation, tone and modal, it investigatesthe possibility of text sentiment analysis. Chapter III and Chapter IV respectively from theword level and sentence level analysis the sentiment orientation. It mainly discusses themodal adverb, negative adverb, adverbs of degree, interrogative pronouns, demonstrativepronouns, and parts of special sentences, which play an important role in the sentimentanalysis. In this chapter we find127emotional expressions which are contained by theabove-mentioned language units, we also list the examples followed by the emotionalexpressions. Chapter V summed the whole work, discussed the meaning of the sentimentanalysis from the perspective of linguistics, and prospected the next work.By examining the corpus, we find the negative emotional expression is closely related tothe negative structures and Bei-sentence. The connection of negative structures and modal adverbs, adverbs of degree, demonstrative pronoun, and interrogative pronouns canstrengthen or weaken the negative effect, through this way, the sentence can express differentdegrees of negative emotional tendency. In contrast, the positive emotional expression mainlyconnects the commendatory term or the negative form of the derogatory term. While thepolarity of a word is usually influenced by the context. So the identification of positivetendency needs more knowledge of context than the identification the negative tendency. Thispaper set the special vocabulary of the mobile phone in order to give the emotional expressiondetailed description.