Unsupervised Learning of Semantic Orientation from a Hundred-Billion-Word Corpus

TL;DR

This paper presents an unsupervised method for determining the semantic orientation of words using large web corpora and PMI analysis, achieving 80% accuracy across various parts of speech.

Contribution

It introduces a simple, scalable algorithm that leverages web search engine queries and PMI to learn semantic orientation without supervision.

Findings

Achieved 80% accuracy on 3,596 words

Effective across adjectives, adverbs, nouns, and verbs

Comparable to supervised methods for adjectives

Abstract

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.