UsingNaYveBayesClassifier toDistinguish Reviews fromNon-review Documents inChinese
Zhang Zi-qiong · 2007
Reviewsaresubjective documents expressing opinions or evaluations. In contrast, non-review documents often present factual information objectively. Separating reviews fromnon-reviews, or subjectivity classification, ispotentially important for manytext processing applications, suchasinformation extraction andinformation retrieval. Also, itisakey process insentiment classification foronline customer reviews. As a typeof genreclassification, the classifications ofsubjective andobjective textsare different fromtraditional topic-based classifications. Not manystudies havebeenconducted inthis domainand mostofthemwereonEnglish texts. Little workhasbeen doneonChinese subjectivity classification. However, the detailed techniques usedinEnglish texts cannotbe applied directly to Chinese dueto thedifferent characteristics between these twolanguages. Thispaper proposesan approach to performsubjectivity classification onChinese textbasedona supervised machine learning algorithm, NaiveBayes. Experiment studies havebeenconducted ontwokinds ofdocuments: moviereviews andmovieplots written inChinese. The results showthattheperformances oftheproposed approach arecomparable tothose oftheexisting English subjectivity classification studies.