Deeper sentiment analysis using machine translation technology

Abstract

This paper proposes a new paradigm for sentiment analysis: translation from text documents to a set of sentiment units. The techniques of deep language analysis for machine translation are applicable also to this kind of text mining task. We developed a high-precision sentiment analysis system at a low development cost, by making use of an existing transfer-based machine translation engine.