title: Unsupervised Learning of Semantic Orientation from a Hundred-Billion-Word Corpus creator: Turney, Peter D. creator: Littman, Michael L. subject: Artificial Intelligence subject: Language subject: Machine Learning subject: Statistical Models description: The evaluative character of a word is called its semantic orientation. A positive semantic orientation implies desirability (e.g., "honest", "intrepid") and a negative semantic orientation implies undesirability (e.g., "disturbing", "superfluous"). This paper introduces a simple algorithm for unsupervised learning of semantic orientation from extremely large corpora. The method involves issuing queries to a Web search engine and using pointwise mutual information to analyse the results. The algorithm is empirically evaluated using a training corpus of approximately one hundred billion words — the subset of the Web that is indexed by the chosen search engine. Tested with 3,596 words (1,614 positive and 1,982 negative), the algorithm attains an accuracy of 80%. The 3,596 test words include adjectives, adverbs, nouns, and verbs. The accuracy is comparable with the results achieved by Hatzivassiloglou and McKeown (1997), using a complex four-stage supervised learning algorithm that is restricted to determining the semantic orientation of adjectives. date: 2002 type: Departmental Technical Report type: NonPeerReviewed format: application/postscript identifier: http://cogprints.org/2322/1/ERB-1094.ps format: application/pdf identifier: http://cogprints.org/2322/5/ERB-1094.pdf identifier: Turney, Peter D. and Littman, Michael L. (2002) Unsupervised Learning of Semantic Orientation from a Hundred-Billion-Word Corpus. [Departmental Technical Report] (Unpublished) relation: http://cogprints.org/2322/