The evaluative character of a word is called its semantic orientation.Positive semantic orientation indicates praise (e.g., "honest", "intrepid") andnegative semantic orientation indicates criticism (e.g., "disturbing","superfluous"). Semantic orientation varies in both direction (positive ornegative) and degree (mild to strong). An automated system for measuringsemantic orientation would have application in text classification, textfiltering, tracking opinions in online discussions, analysis of surveyresponses, and automated chat systems (chatbots). This paper introduces amethod for inferring the semantic orientation of a word from its statisticalassociation with a set of positive and negative paradigm words. Two instancesof this approach are evaluated, based on two different statistical measures ofword association: pointwise mutual information (PMI) and latent semanticanalysis (LSA). The method is experimentally tested with 3,596 words (includingadjectives, adverbs, nouns, and verbs) that have been manually labeled positive(1,614 words) and negative (1,982 words). The method attains an accuracy of82.8% on the full test set, but the accuracy rises above 95% when the algorithmis allowed to abstain from classifying mild words.
translated by 谷歌翻译