Learning to disambiguate potentially subjective expressions

The goal of this work is recognizing opinionated and evaluative (subjective) language in text. The ability to recognize such language would be beneficial for many NLP applications such as question answering, information extraction, summarization, and genre detection. This paper focuses on disambiguating potentially subjective expressions in context, based on the density of other clues in the surrounding text.