Learning Affective Language and Its Application

Abstract

Affective in natural language refers to aspects of language used to express opinions, emotions, and beliefs. There are numerous natural language processing applications for which affective analysis is relevant, including online chat, news comment, predicting of the stock market and tracking customers’ emotion states. The goal of this work is learning affective language from corpora and using this knowledge for affective analysis. Clues of affective are generated and tested, including unique words, collocations based on dependency grammar, and composition feature using distributional semantic models. The clues, generated from different data sets using different procedures, then the clues are used to perform affective analysis to demonstrate the utility of the knowledge acquired in this paper.